{"format":"fdedictionary-list","version":1,"exported_at":"2026-07-21T16:47:50+00:00","entries":[{"term":"acceptance criteria","term_normalized":"acceptance criteria","senses":[{"definition":"Specific conditions the customer and FDE agree must be true before a feature, workflow, or deployment is considered done.","usage":"The acceptance criteria required 95 percent of cases to include the right source citation."}]},{"term":"action button","term_normalized":"action button","senses":[{"definition":"A UI control that triggers a real workflow step: approval, escalation, submission, write-back, or agent tool use.","usage":"The FDE added an action button so supervisors could approve the drafted update."}]},{"term":"actionability","term_normalized":"actionability","senses":[{"definition":"The degree to which output gives a user enough specific, trustworthy information to take the next step. An answer can be accurate but still lack actionability.","usage":"The answer was accurate but lacked actionability because it didn't say who should approve it."}]},{"term":"adoption blocker","term_normalized":"adoption blocker","senses":[{"definition":"Anything that keeps users or sponsors from using the deployment in normal work: trust, data quality, UX friction, unclear ownership, policy, incentives, or a missing integration. Usually not accuracy.","usage":"The adoption blocker was not accuracy; it was that users had to copy the answer into another system."}]},{"term":"adoption loop","term_normalized":"adoption loop","senses":[{"definition":"A repeatable cycle: ship, watch real usage, collect feedback, remove friction, expand. Run it weekly until the workflow is boring and reliable.","usage":"The FDE ran the adoption loop weekly until the workflow became boring and reliable."}]},{"term":"adoption path","term_normalized":"adoption path","senses":[{"definition":"The sequence from first champion to durable production usage. For example: pilots, training, owner handoff, metrics, and expansion.","usage":"The adoption path started with five reviewers and ended with the queue becoming mandatory."}]},{"term":"adoption playbook","term_normalized":"adoption playbook","senses":[{"definition":"A reusable set of tactics for driving usage: champion prep, training, launch comms, metrics, office hours, escalation paths. Updated with each deployment.","usage":"The FDE updated the adoption playbook after learning supervisors needed a separate walkthrough."}]},{"term":"adoption workflow","term_normalized":"adoption workflow","senses":[{"definition":"A workflow design that makes the new capability part of normal work instead of an optional side tool users have to remember to visit.","usage":"The adoption workflow put the AI summary inside the ticket review screen."}]},{"term":"agent","term_normalized":"agent","senses":[{"definition":"An AI system that uses context and tools to perform steps in a workflow under defined constraints, rather than just answering a question.","usage":"The agent summarized the case, checked policy, and drafted the response."}]},{"term":"agent guardrail","term_normalized":"agent guardrail","senses":[{"definition":"A control that limits what an agent can see, say, decide, or do, which is enforced through policy, permissions, validation, or runtime checks.","usage":"The agent guardrail blocked write-back unless the user approved the change."}]},{"term":"agent handoff","term_normalized":"agent handoff","senses":[{"definition":"The designed transfer of work between an agent and a human, another agent, or a system, which is triggered when confidence, ownership, or permissions change.","usage":"The agent handoff sent low-confidence cases to a supervisor with the evidence attached."}]},{"term":"agent operating model","term_normalized":"agent operating model","senses":[{"definition":"The technical and organizational model for building, approving, monitoring, supporting, escalating, and improving agents after launch.","usage":"The agent operating model assigned support ownership before rollout."}]},{"term":"agent orchestration","term_normalized":"agent orchestration","senses":[{"definition":"The logic that coordinates agents, tools, prompts, retrieval, state, and human checkpoints across a workflow. FDEs reduce this until it's as simple as the use case allows.","usage":"The FDE reduced agent orchestration to one router and three tools."}]},{"term":"agent rollout","term_normalized":"agent rollout","senses":[{"definition":"The staged introduction of an agent to users, workflows, permissions, or action types. Recommendations before write-back is a common first stage.","usage":"The agent rollout started with recommendations before write-back."}]},{"term":"agent skill","term_normalized":"agent skill","senses":[{"definition":"A bounded capability an agent can perform, which is defined by instructions, tools, inputs, outputs, permissions, and evals.","usage":"The FDE added an agent skill for drafting renewal summaries."}]},{"term":"agent workflow","term_normalized":"agent workflow","senses":[{"definition":"The ordered steps an agent follows: context gathering, reasoning, tool calls, decisions, human checkpoints, and outputs. FDEs simplify these after watching them in practice.","usage":"The FDE simplified the agent workflow after observing three unnecessary tool calls."}]},{"term":"agentic application","term_normalized":"agentic application","senses":[{"definition":"An application where AI agents perform meaningful workflow steps through tools, context, and controls (not just generating text for a user to copy somewhere else).","usage":"The agentic application opened the right claim view and prepared the adjustment recommendation."}]},{"term":"agentic enterprise","term_normalized":"agentic enterprise","senses":[{"definition":"An enterprise operating model where agents participate in many workflows under shared governance, monitoring, and integration patterns. Needs standard tool permissions before teams build dozens of agents independently.","usage":"The agentic enterprise needed standard tool permissions before teams built dozens of agents."}]},{"term":"agentic workflow","term_normalized":"agentic workflow","senses":[{"definition":"A workflow where an AI agent plans and performs multi-step work using tools, context, and controls (rather than just returning a chat response).","usage":"The agentic workflow retrieved the case, checked policy, drafted the response, and waited for approval."}]},{"term":"AGI","term_normalized":"agi","senses":[{"definition":"See: artificial general intelligence"}]},{"term":"AI adoption","term_normalized":"ai adoption","senses":[{"definition":"Getting users, operators, and sponsors to rely on AI in real work (not just approving a pilot and forgetting it).","usage":"The FDE measured AI adoption by completed reviews, not demo attendance."}]},{"term":"AI adoption engineer","term_normalized":"ai adoption engineer","senses":[{"definition":"An engineer focused on making AI actually used: fitting it into existing workflows, closing trust gaps, training users, and measuring whether the deployment changes how the operation actually runs.","usage":"The AI adoption engineer moved the assistant from a side tab into the review step users already followed."}]},{"term":"AI agent deployment","term_normalized":"ai agent deployment","senses":[{"definition":"Launching an agent that can use customer context and tools under defined permissions, guardrails, evals, observability, and human handoffs — not just prompting a model in production.","usage":"The AI agent deployment started in read-only mode before allowing approved write-back."}]},{"term":"AI deployment engineer","term_normalized":"ai deployment engineer","senses":[{"definition":"An engineer responsible for getting AI features or agents safely into live customer use — integrations, evals, access controls, monitoring, and user enablement all included.","usage":"The AI deployment engineer blocked launch until audit logs and fallback handling were in place."}]},{"term":"AI deployment playbook","term_normalized":"ai deployment playbook","senses":[{"definition":"A deployment playbook specific to AI systems — covering use-case selection, data access, evals, model behavior, guardrails, monitoring, rollout, and adoption.","usage":"The AI deployment playbook required a shadow-mode phase for any write-capable agent."}]},{"term":"AIP","term_normalized":"aip","senses":[{"definition":"Palantir's Artificial Intelligence Platform: a governed environment for building AI applications and workflows over enterprise data, tools, and ontology-backed permissions.","usage":"The FDE used AIP to connect governed data, an approval workflow, and an AI assistant."}]},{"term":"AIP Bootcamp","term_normalized":"aip bootcamp","senses":[{"definition":"A Palantir-style intensive workshop where FDEs and customer teams identify use cases, build working AI workflows on AIP, and define the path to production — often over a few days onsite.","usage":"The AIP Bootcamp produced three prototypes and one workflow owner willing to take the first one live."}]},{"term":"air-gapped deployment","term_normalized":"air-gapped deployment","senses":[{"definition":"A deployment in an isolated environment with no or tightly controlled external network connectivity — changes how the FDE packages updates, credentials, and model access.","usage":"The air-gapped deployment changed how the FDE packaged model updates."}]},{"term":"applied AI engineer","term_normalized":"applied ai engineer","senses":[{"definition":"An engineer who turns AI capabilities into working customer applications by combining models, tools, data, evals, product code, and rollout support. Builds with real data, not synthetic examples.","usage":"The applied AI engineer built the first version using real support tickets instead of made-up prompts."}]},{"term":"applied AI FDE","term_normalized":"applied ai fde","senses":[{"definition":"A forward deployed engineer focused specifically on applying AI to customer workflows, owning the full arc: discovery, prototyping, evaluation, integration, and adoption.","usage":"The applied AI FDE found that the hard part wasn't the model call — it was the human approval step after it."}]},{"term":"approval gate","term_normalized":"approval gate","senses":[{"definition":"A required approval step before a workflow can proceed or a system action can be taken.","usage":"The approval gate stopped the agent from sending customer emails directly."}]},{"term":"artificial general intelligence","term_normalized":"artificial general intelligence","senses":[{"definition":"A machine that can do anything that people can do, better than people.","usage":"The artificial general intelligence can do anything better than people."},{"definition":"An AI that you can just tell to figure out how to make money and do it profitably for you, and it figures it out and does it on its own."},{"definition":"A mostly automatic system that does better than people at a majority of commercially valuable roles.","usage":"OpenAI’s mission is to ensure that artificial general intelligence (AGI)—by which we mean highly autonomous systems that outperform humans at most economically valuable work—benefits all of humanity. --OpenAI Charter at https://openai.com/charter"},{"definition":"Any technology that seems smart enough that it seems to be intelligent in a general way, meaning it can do different things once only doable by humans, much better than most humans who are not expert in that one field can do that one activity (such as: write essays or literature, summarize documents, produce reports and analyses, create music, create images, code software, scrutinize research, suggest strategies, transcribe audio to text, translate text among arbitrary languages, converse fluently over spoken words, or hold a detailed an in-depth spoken dialog on just about any topic in the world at far above an average human level of knowledge for any given topic).\r\n\r\nSuch a technology is also typically much faster than even humans highly skilled in that field can doing its activities.\r\n\r\nThe technology's output may match the output quality of humans in many of its tasks.  The technology's output is not necessarily at a higher quality than that of all humans highly skilled in that field who spend a much larger amount of time on the same exact task.   However, it is quite possible that the technology's output is at higher quality than that of many highly skilled, skilled, or semi-skilled humans who spend a much larger amount of time on the same task."},{"definition":"A system that can understand, learn, and apply knowledge across a wide range of tasks at a level equal to or exceeding human intelligence."}]},{"term":"audit trail","term_normalized":"audit trail","senses":[{"definition":"A durable record of what happened, who or what did it, what data was used, and when — required before any agent touches regulated records.","usage":"The audit trail was required before the agent could touch regulated records."}]},{"term":"before-and-after workflow","term_normalized":"before-and-after workflow","senses":[{"definition":"A comparison of how work happened before the deployment and how it happens after — the clearest way to communicate value to non-technical stakeholders.","usage":"The before-and-after workflow made the ROI obvious to finance."}]},{"term":"bespoke agentic solution","term_normalized":"bespoke agentic solution","senses":[{"definition":"A customer-specific agentic solution built for a particular workflow — useful as a first deployment, but needs to be evaluated for what belongs in product before the second account asks for the same thing.","usage":"The FDE shipped a bespoke agentic solution, then identified which parts belonged in product."}]},{"term":"blast radius","term_normalized":"blast radius","senses":[{"definition":"The potential scope of damage if a deployment, agent, credential, or integration fails. FDEs reduce blast radius by limiting write-back scope, using staged rollout, and enforcing least privilege.","usage":"The FDE reduced blast radius by limiting write-back to one region."}]},{"term":"builder-consultant","term_normalized":"builder-consultant","senses":[{"definition":"A hybrid role that combines consultant-style problem framing with engineering ownership of code, debugging, and shipment. An FDE plays this role constantly.","usage":"The FDE acted as a builder-consultant: scoped the workflow with the VP, then built the connector."}]},{"term":"business outcome","term_normalized":"business outcome","senses":[{"definition":"The customer result the deployment is meant to improve — stated in operational or financial terms. The thing FDEs tie every technical decision back to.","usage":"The business outcome was fewer missed renewals, not a better chat experience."}]},{"term":"business-critical workflow","term_normalized":"business-critical workflow","senses":[{"definition":"A workflow where failures affect revenue, service levels, compliance, safety, or executive attention. The FDE treats it like production from day one.","usage":"The FDE treated the business-critical workflow like production from day one."}]},{"term":"canary rollout","term_normalized":"canary rollout","senses":[{"definition":"A small initial rollout to a limited user group or workflow slice to catch issues before the broader release sees them.","usage":"The canary rollout exposed a permissions bug before the national team saw it."}]},{"term":"change management","term_normalized":"change management","senses":[{"definition":"The work of helping an organization adopt a new workflow through communication, training, incentives, support, and leadership alignment. An FDE treats this as part of deployment, not the customer's problem.","usage":"The FDE treated change management as part of deployment, not a customer afterthought."}]},{"term":"change window","term_normalized":"change window","senses":[{"definition":"The approved time period when production changes may be made in a customer environment.","usage":"The FDE scheduled cutover during the customer's change window."}]},{"term":"citation requirement","term_normalized":"citation requirement","senses":[{"definition":"A product or workflow requirement that model answers cite the underlying sources used — usually driven by legal, compliance, or user trust needs.","usage":"The citation requirement came from legal, not engineering."}]},{"term":"client-embedded engineer","term_normalized":"client-embedded engineer","senses":[{"definition":"An engineer placed close enough to a client team that discovery, build, testing, and rollout all happen against the client's actual operating process.","usage":"The client-embedded engineer learned the real bottleneck by sitting with the reviewers for a day."}]},{"term":"co-design","term_normalized":"co-design","senses":[{"definition":"Designing the solution with the customer's users and technical owners while the workflow is still being built — so the product fits the real operation, not an imagined one.","usage":"The FDE used co-design with supervisors to decide where the agent should stop and ask for approval."}]},{"term":"compliance evidence","term_normalized":"compliance evidence","senses":[{"definition":"Artifacts proving the deployment meets required policy, security, legal, or regulatory controls — logs, diagrams, access reviews, and test results.","usage":"The FDE gathered compliance evidence from logs, diagrams, and access reviews."}]},{"term":"confidence threshold","term_normalized":"confidence threshold","senses":[{"definition":"A threshold used to decide when model or agent output is good enough to proceed, requires human review, or should fall back to a simpler path.","usage":"The FDE raised the confidence threshold for regulated cases."}]},{"term":"connective tissue","term_normalized":"connective tissue","senses":[{"definition":"The technical and organizational glue linking customer users, product teams, data owners, engineers, and deployment milestones into one executable path. Usually an FDE.","usage":"The FDE was connective tissue between security, product, and the operations team."}]},{"term":"core-product feedback loop","term_normalized":"core-product feedback loop","senses":[{"definition":"A feedback loop where deployment lessons become changes to the core product instead of staying as customer-specific patches.","usage":"The core-product feedback loop moved custom role mapping into the platform."}]},{"term":"cost envelope","term_normalized":"cost envelope","senses":[{"definition":"The acceptable cost range for running a deployment — model usage, infrastructure, support, and integration maintenance. Shapes model routing and caching decisions.","usage":"The cost envelope made the team use a cheaper model for classification."}]},{"term":"customer pain point","term_normalized":"customer pain point","senses":[{"definition":"A specific operational friction or failure painful enough that the customer will change workflow, assign owners, and support engineering work to fix it.","usage":"The customer pain point was the manual reconciliation between the billing system and CRM."}]},{"term":"customer sandbox","term_normalized":"customer sandbox","senses":[{"definition":"A customer-controlled test environment used to validate integrations, permissions, and workflows before production. Usually has fake or masked data and less strict network rules than prod.","usage":"The customer sandbox had fake data, so the FDE still needed a production dry run."}]},{"term":"customer workaround","term_normalized":"customer workaround","senses":[{"definition":"A manual or improvised process customers use when the product doesn't fit their workflow. The best source of product gaps if the FDE takes time to document it.","usage":"The customer workaround was a spreadsheet that had become the source of truth."}]},{"term":"customer-facing engineer","term_normalized":"customer-facing engineer","senses":[{"definition":"An engineer expected to work directly with customers while still writing, debugging, and reviewing real implementation work. Not advisory — owns code.","usage":"The customer-facing engineer explained the tradeoff to the VP, then opened the pull request."}]},{"term":"cutover","term_normalized":"cutover","senses":[{"definition":"The switch from the old process or system path to the new deployed one.","usage":"The cutover happened after the last batch job completed."}]},{"term":"data contract","term_normalized":"data contract","senses":[{"definition":"A clear agreement about the schema, semantics, freshness, quality, and ownership of data flowing into a deployment. Without one, the deployment breaks silently when the source changes.","usage":"The deployment kept breaking until the FDE wrote a data contract with the warehouse team."}]},{"term":"data owner","term_normalized":"data owner","senses":[{"definition":"The customer person or team accountable for access, quality, meaning, and governance of a data source. Their approval is required before that data enters a deployment.","usage":"The data owner approved the policy corpus for retrieval."}]},{"term":"degraded mode","term_normalized":"degraded mode","senses":[{"definition":"A reduced-function mode used when a dependency is unavailable but the workflow can still provide partial value safely — show prior summaries, disable write actions.","usage":"In degraded mode, the assistant showed prior summaries but disabled write actions."}]},{"term":"dependency map","term_normalized":"dependency map","senses":[{"definition":"A map of systems, teams, data sources, credentials, jobs, and approvals the deployment depends on — useful for launch planning and blast radius analysis.","usage":"The dependency map revealed a batch job owned by finance could delay launch."}]},{"term":"deployment","term_normalized":"deployment","senses":[{"definition":"The full process of making a capability work for real users in a real environment: configuration, integration, validation, launch, and support. Not done until the support team knows how to handle failures.","usage":"The FDE considered deployment unfinished until the support team knew how to handle failures."}]},{"term":"deployment blocker","term_normalized":"deployment blocker","senses":[{"definition":"A launch-blocking issue in the deployment path — often not an engineering problem but unclear business ownership, missing approval, or an unresolved data issue.","usage":"The deployment blocker was not engineering effort; it was unclear business ownership."}]},{"term":"deployment engineer","term_normalized":"deployment engineer","senses":[{"definition":"An engineer who owns the practical work of launching software in a customer environment: configuration, integration, testing, troubleshooting, and handoff to support.","usage":"The deployment engineer found the launch blocker in the customer's SSO group mapping."}]},{"term":"deployment narrative","term_normalized":"deployment narrative","senses":[{"definition":"The simple story that explains what was built, why it matters, who uses it, and what result it produced — for the sponsor, not the engineering team.","usage":"The FDE sharpened the deployment narrative before the sponsor presented it internally."}]},{"term":"deployment pattern","term_normalized":"deployment pattern","senses":[{"definition":"A repeatable way to get a class of customer use cases into production — architecture, rollout, common blockers, support model — documented after the third similar launch.","usage":"The FDE documented the deployment pattern after the third similar launch."}]},{"term":"deployment playbook","term_normalized":"deployment playbook","senses":[{"definition":"A practical guide for repeating a deployment: prerequisites, architecture, steps, blockers, owners, tests, launch plan, and support model. Lets another FDE run the same play without rediscovering everything.","usage":"The deployment playbook helped a new FDE launch the same workflow at another account."}]},{"term":"deployment pod","term_normalized":"deployment pod","senses":[{"definition":"A cross-functional team responsible for taking one customer deployment or deployment pattern to production.","usage":"The deployment pod owned the backlog, launch checklist, and first-week support."}]},{"term":"deployment wedge","term_normalized":"deployment wedge","senses":[{"definition":"The first narrow but valuable workflow that creates trust, proves value, and opens the door to broader deployment. Should be painful to the customer, easy to measure, and winnable.","usage":"The deployment wedge was invoice triage because it was painful and easy to measure."}]},{"term":"design partner","term_normalized":"design partner","senses":[{"definition":"A customer who works closely with the product or field team to shape a capability before it is broadly available — accepting rough edges in exchange for influence over the roadmap.","usage":"The design partner accepted rough edges in exchange for influence over the roadmap."}]},{"term":"domain object","term_normalized":"domain object","senses":[{"definition":"A business entity meaningful to the customer — a claim, account, order, patient, asset, invoice, or case — that should anchor the workflow rather than a database row.","usage":"The claim became the domain object that anchored the workflow."}]},{"term":"dry run","term_normalized":"dry run","senses":[{"definition":"A test execution that exercises the full workflow without committing production changes.","usage":"The FDE did a dry run of the renewal workflow before enabling write-back."}]},{"term":"edge case","term_normalized":"edge case","senses":[{"definition":"A less common but important scenario that can break trust, safety, or usefulness if the deployment mishandles it. FDEs find these by watching real users, not by reading requirements.","usage":"The edge case was a customer with two active policies."}]},{"term":"embedded engineer","term_normalized":"embedded engineer","senses":[{"definition":"In FDE context: an engineer temporarily placed inside a customer team to build with their real users, tools, and constraints rather than from a safe distance.","usage":"The embedded engineer joined dispatch standup to learn why the proposed workflow would fail on night shift."}]},{"term":"enterprise AI deployment","term_normalized":"enterprise ai deployment","senses":[{"definition":"Deploying AI into an enterprise environment with production controls: SSO, permissions, audit, data governance, monitoring, support, and change management all required.","usage":"The enterprise AI deployment needed security review before users could upload customer data."}]},{"term":"enterprise-grade agent","term_normalized":"enterprise-grade agent","senses":[{"definition":"An agent built for enterprise use: permissioned, auditable, evaluated, monitored, with fallback paths and human handoffs — not a demo agent in a production slot.","usage":"The enterprise-grade agent could not call a write tool without human approval."}]},{"term":"environment parity","term_normalized":"environment parity","senses":[{"definition":"How closely staging, sandbox, and production match in data shape, permissions, network rules, and system behavior. Poor parity is the root cause of most 'worked in staging' failures.","usage":"Poor environment parity explained why the test passed and production failed."}]},{"term":"escalation path","term_normalized":"escalation path","senses":[{"definition":"The defined route for raising an issue to the right person or team when the deployment cannot proceed normally.","usage":"The escalation path sent tool failures to the platform owner, not the business sponsor."}]},{"term":"escape hatch","term_normalized":"escape hatch","senses":[{"definition":"A deliberate manual override or alternate path that lets operators recover when automation, data, or integrations fail — should be designed in, not bolted on after launch.","usage":"The escape hatch let supervisors take over the case from the agent."}]},{"term":"eval harness","term_normalized":"eval harness","senses":[{"definition":"The test infrastructure used to run model or agent evaluations repeatedly against examples, expected behavior, tools, and scoring logic. FDEs add real failure cases from production.","usage":"The FDE added the customer's top failure cases to the eval harness."}]},{"term":"eval rubric","term_normalized":"eval rubric","senses":[{"definition":"The scoring criteria used to judge whether model or agent output is good enough for the use case — defined by what the customer actually cares about, not by fluency.","usage":"The eval rubric penalized answers without cited policy sections."}]},{"term":"evaluation dataset","term_normalized":"evaluation dataset","senses":[{"definition":"A set of representative examples used to measure model, agent, retrieval, or workflow performance before and after changes.","usage":"The evaluation dataset came from real tickets, stripped of sensitive fields."}]},{"term":"event-driven integration","term_normalized":"event-driven integration","senses":[{"definition":"An integration pattern where changes in one system emit events that trigger downstream workflows or updates in near real time.","usage":"The FDE used event-driven integration so new claims entered the review queue automatically."}]},{"term":"executive sponsor map","term_normalized":"executive sponsor map","senses":[{"definition":"A map of the customer leaders who can fund, unblock, mandate, or kill the deployment — and what each one actually cares about.","usage":"The executive sponsor map showed security approval mattered more than the COO's enthusiasm."}]},{"term":"fail open","term_normalized":"fail open","senses":[{"definition":"A design where the system continues when a check fails. Sometimes acceptable for low-stakes actions; dangerous for regulated or write-capable workflows.","usage":"The team rejected fail open for regulated approvals."}]},{"term":"fallback path","term_normalized":"fallback path","senses":[{"definition":"The designed path when automation cannot proceed safely — routing to a human, using a simpler workflow, or reverting to manual processing.","usage":"The fallback path sent ambiguous cases to the existing triage queue."}]},{"term":"FDE","term_normalized":"fde","senses":[{"definition":"See: forward deployed engineer"},{"definition":"Forward Deployed Engineer. An engineer who lives at the customer, writes real code, integrates systems, and owns the path from ambiguous use case to working production deployment.","usage":"The FDE spent the morning with claims ops and the afternoon shipping the API fix."}]},{"term":"FDE manager","term_normalized":"fde manager","senses":[{"definition":"A manager who staffs FDE work, protects technical quality, unblocks customer situations, and turns field patterns into repeatable deployment improvements and product inputs.","usage":"The FDE manager pushed back on a custom build that should have become a platform feature."}]},{"term":"FDE pod","term_normalized":"fde pod","senses":[{"definition":"A small field team assigned to a customer, segment, or deployment motion with shared ownership for discovery, build, rollout, and adoption.","usage":"The FDE pod met daily until the write-back integration was stable."}]},{"term":"field deployment","term_normalized":"field deployment","senses":[{"definition":"A deployment delivered in close collaboration with the customer environment — not generic self-serve configuration. Usually requires custom data mapping, integration work, or workflow design.","usage":"The field deployment required custom data mapping for the customer's asset hierarchy."}]},{"term":"field insight","term_normalized":"field insight","senses":[{"definition":"A lesson from real customer work that should influence product, documentation, deployment patterns, or strategy — more valuable than a survey because it comes from someone who built it.","usage":"The field insight was that every customer wanted approval chains, not just chat."}]},{"term":"field-led deployment","term_normalized":"field-led deployment","senses":[{"definition":"A deployment motion where hands-on FDE work is necessary because the customer environment, workflow, or value proposition is too complex for self-serve.","usage":"The account needed field-led deployment because the workflow spanned four systems."}]},{"term":"forward deployed AI engineer","term_normalized":"forward deployed ai engineer","senses":[{"definition":"An FDE focused on AI systems: agents, model behavior, retrieval pipelines, tool integrations, evals, permissions, and the workflow redesign needed to make AI actually useful in production.","usage":"The forward deployed AI engineer tuned the agent only after watching how analysts handled exceptions manually."}]},{"term":"forward deployed engineer","term_normalized":"forward deployed engineer","senses":[{"definition":"Aka FDE.  An individual who helps an org get the most out of AI capabilities by rapidly converting discovered intelligence patterns into software.  For AI tools, they identify where that org uses intelligence and delivers on the AI promise.\r\n\r\nAn FDE\r\n  (1) audits to see how to rebuild from the ground up around a particular product or AI;\r\n  (2) sets up evaluations to make sure the automation/AI-use will be at least as good as humans; and\r\n  (3) deploy that so the systems are real and stay good.\r\n\r\nTo do this, the FDE is at the intersection of communication, intelligence, product, engineering, and AI.  They effectively apply diplomatic pragmatism, effectively selling what they are doing based on their ability to scrutinize and quickly identify, create, deliver, and articulate the value they deliver.","usage":"That forward deployed engineer (FDE) does what might have taken a team of 2-5 in the past.  But at a fraction of what such a team would cost today.  Which would be hard to do efficiently with a team much larger than that.  With a solution that would have been impossible a few years ago.  I guess that's why we brought them in."},{"definition":"An engineer embedded with customers to discover, build, integrate, debug, and productionize software or AI inside the customer's real environment. Not there to write decks — there to make the thing work where the customer actually works.","usage":"A forward deployed engineer is not there to make slides; they are there to make the thing work."}]},{"term":"forward deployed engineering manager","term_normalized":"forward deployed engineering manager","senses":[{"definition":"A leader for FDE teams who balances customer outcomes, engineering execution, staffing, escalation, and the field-to-product feedback loop.","usage":"The forward deployed engineering manager moved an engineer onto the account when the rollout became production-critical."}]},{"term":"forward deployed software engineer","term_normalized":"forward deployed software engineer","senses":[{"definition":"An FDE whose primary work is building and integrating software in customer environments — product code, data pipelines, APIs, UX — wherever the standard connector doesn't fit.","usage":"The forward deployed software engineer owned the integration because the standard connector didn't match the customer's data model."}]},{"term":"Frontier Alliance","term_normalized":"frontier alliance","senses":[{"definition":"OpenAI's partner program for helping enterprises deploy frontier AI through trained service and consulting partners.","usage":"The Frontier Alliance model created more demand for repeatable FDE playbooks."}]},{"term":"go-live support","term_normalized":"go-live support","senses":[{"definition":"Hands-on FDE support around launch: monitoring, triage, fixes, user help, rollback decisions, and escalation management. The FDE stays until the workflow is stable.","usage":"The FDE stayed in go-live support until the first hundred cases processed cleanly."}]},{"term":"golden dataset","term_normalized":"golden dataset","senses":[{"definition":"A curated set of examples with trusted expected outcomes — used to test model or workflow behavior and catch regressions. Built from real cases, not invented ones.","usage":"The golden dataset included the weird edge cases operators cared about."}]},{"term":"grounded answer","term_normalized":"grounded answer","senses":[{"definition":"An answer tied to source data, documents, or tool results that users can inspect or verify — not a confident-sounding guess.","usage":"The FDE required a grounded answer for every compliance recommendation."}]},{"term":"guardrail bypass","term_normalized":"guardrail bypass","senses":[{"definition":"A failure mode where a user, prompt, tool, or configuration avoids a control that was supposed to limit behavior. Treated as a launch-blocking bug.","usage":"The FDE treated the guardrail bypass as a launch-blocking bug."}]},{"term":"happy path","term_normalized":"happy path","senses":[{"definition":"The straightforward case where the workflow works as expected, without edge cases, missing data, or integration failures. Demos run on the happy path; production does not.","usage":"The happy path demo worked, but the FDE still needed to test exceptions."}]},{"term":"human checkpoint","term_normalized":"human checkpoint","senses":[{"definition":"A specific point in a workflow where a human must review, approve, resolve ambiguity, or accept responsibility before the workflow continues.","usage":"The FDE added a human checkpoint before refunds over $500."}]},{"term":"human-agent workflow","term_normalized":"human-agent workflow","senses":[{"definition":"A workflow where humans and AI agents share the work, with clear ownership, review points, escalation paths, and audit trails for when each side holds the ball.","usage":"The FDE designed the human-agent workflow so the agent drafted and the supervisor approved."}]},{"term":"implementation engineer","term_normalized":"implementation engineer","senses":[{"definition":"An engineer who adapts, configures, integrates, and tests a product for a specific customer environment until it meets the agreed acceptance criteria.","usage":"The implementation engineer rewired the import job around the customer's actual account hierarchy."}]},{"term":"implementation strategy","term_normalized":"implementation strategy","senses":[{"definition":"The plan for sequencing technical work, customer decisions, risks, rollout, and handoff so implementation reaches production efficiently.","usage":"The implementation strategy shipped read-only retrieval first and deferred automation."}]},{"term":"integration design","term_normalized":"integration design","senses":[{"definition":"The plan for connecting systems, data, identity, tools, and workflows so the deployed capability works reliably — and keeps working after the FDE leaves.","usage":"The integration design kept retrieval separate from write-back."}]},{"term":"integration surface","term_normalized":"integration surface","senses":[{"definition":"The set of APIs, data objects, credentials, events, permissions, and systems touched by an integration. FDEs try to minimize this for the first launch.","usage":"The FDE narrowed the integration surface to reduce launch risk."}]},{"term":"land and expand","term_normalized":"land and expand","senses":[{"definition":"A growth motion where a small, successful deployment creates trust and evidence for broader use across the customer account.","usage":"The deployment wedge supported a land and expand motion into adjacent workflows."}]},{"term":"latency budget","term_normalized":"latency budget","senses":[{"definition":"The total time a workflow can spend on model calls, retrieval, tool calls, and UI updates before users lose patience or trust. Forces caching and architectural choices.","usage":"The latency budget forced the FDE to cache policy snippets."}]},{"term":"launch checklist","term_normalized":"launch checklist","senses":[{"definition":"A checklist of everything that must be true before go-live: owners named, access confirmed, evals passed, tests done, monitoring active, comms sent, support ready, and fallbacks in place.","usage":"The launch checklist caught that nobody had trained the night-shift supervisors."}]},{"term":"least privilege","term_normalized":"least privilege","senses":[{"definition":"Giving users, services, and agents only the permissions needed to do their job — and no more. Applied to service accounts, tool scopes, and agent permissions.","usage":"The FDE applied least privilege to the service account from the start."}]},{"term":"live deployment","term_normalized":"live deployment","senses":[{"definition":"A deployment currently used by real users against production or production-like systems.","usage":"The live deployment broke when the customer's identity provider changed group names."}]},{"term":"low-code extension","term_normalized":"low-code extension","senses":[{"definition":"A customer- or field-built extension using low-code configuration wherever possible and code only where necessary.","usage":"The low-code extension handled the simple routing rules without another service."}]},{"term":"manual-process reduction","term_normalized":"manual-process reduction","senses":[{"definition":"Reducing repetitive human steps while preserving the judgment, controls, and accountability that still need a person.","usage":"The FDE measured manual-process reduction by how many copy-paste steps disappeared."}]},{"term":"MCP connector","term_normalized":"mcp connector","senses":[{"definition":"A connector that exposes a specific system, API, or data source through MCP so an AI application can use it safely with standard tool calling.","usage":"The FDE wrote an MCP connector for the internal knowledge base."}]},{"term":"MCP layer","term_normalized":"mcp layer","senses":[{"definition":"The controlled integration layer that exposes customer tools, data, or actions to AI applications through MCP servers — enforcing permissions without baking them into every agent.","usage":"The MCP layer gave the agent access to account data without bypassing permissions."}]},{"term":"MCP server","term_normalized":"mcp server","senses":[{"definition":"A server implementing the Model Context Protocol so AI systems can access tools, resources, or prompts through a standard interface — the preferred integration layer for FDEs building agent connections to customer systems.","usage":"The FDE built an MCP server for the customer's ticketing system."}]},{"term":"model adoption blocker","term_normalized":"model adoption blocker","senses":[{"definition":"A blocker specific to trusting model output: hallucinations, missing citations, brittle prompts, failed evals, or unclear accountability for wrong answers.","usage":"The model adoption blocker disappeared once every answer cited the customer policy it used."}]},{"term":"model demo","term_normalized":"model demo","senses":[{"definition":"A focused demonstration of model behavior on customer-relevant examples to test value, expose gaps, and decide what has to change before deployment. Not a sales deck — a working diagnostic.","usage":"The model demo made it obvious users needed source citations before they would trust any answer."}]},{"term":"monitoring dashboard","term_normalized":"monitoring dashboard","senses":[{"definition":"A dashboard used to track deployment health, adoption, errors, model behavior, or operational outcomes — built before launch, not after the first incident.","usage":"The monitoring dashboard showed support that failures were coming from one downstream API."}]},{"term":"multi-agent workflow","term_normalized":"multi-agent workflow","senses":[{"definition":"A workflow that coordinates more than one agent or specialized model role, usually with explicit routing, tool access, and handoff rules between them.","usage":"The multi-agent workflow separated retrieval, reasoning, and customer-response drafting."}]},{"term":"object-aware application","term_normalized":"object-aware application","senses":[{"definition":"An application that understands customer domain objects, their relationships, permissions, lifecycle, and workflow actions — not just raw data.","usage":"The object-aware application treated every claim as a governed object with history."}]},{"term":"offline eval","term_normalized":"offline eval","senses":[{"definition":"Evaluation run outside live user traffic — on a saved dataset — before changes reach production. Used to catch regressions before users see them.","usage":"The offline eval caught a regression in citation quality."}]},{"term":"on-prem integration","term_normalized":"on-prem integration","senses":[{"definition":"Connecting a cloud or SaaS deployment to systems running in the customer's own data center or managed infrastructure. Usually requires a gateway or proxy because the systems are not internet-facing.","usage":"The on-prem integration required a gateway because the database was not internet-facing."}]},{"term":"one-off customer solution","term_normalized":"one-off customer solution","senses":[{"definition":"A customer-specific solution that solves the immediate problem but won't scale or reuse without refactoring. Acceptable for a pilot; a problem if it ships as the permanent answer.","usage":"The FDE accepted a one-off customer solution for the pilot, then logged the product gap."}]},{"term":"online eval","term_normalized":"online eval","senses":[{"definition":"Evaluation using live or near-live traffic, feedback, or production outcomes — measures what actually happens when users interact with the deployed system.","usage":"The online eval measured whether users accepted the agent's drafts."}]},{"term":"ontology","term_normalized":"ontology","senses":[{"definition":"A structured model of the customer's business objects, relationships, actions, permissions, and semantics — used by Palantir's AIP platform as the foundation for governed workflows.","usage":"The ontology let the workflow reason over assets, work orders, and maintenance events."}]},{"term":"ontology-backed workflow","term_normalized":"ontology-backed workflow","senses":[{"definition":"A workflow built on a governed object model so actions, permissions, and context align with the customer's actual business entities — not arbitrary database rows.","usage":"The ontology-backed workflow let the agent update an asset, not just a row in a table."}]},{"term":"operational debt","term_normalized":"operational debt","senses":[{"definition":"Unowned scripts, brittle workflows, manual steps, or unsupported customizations created during a deployment and left for operators to carry. An FDE should pay this down before handoff.","usage":"The FDE paid down operational debt before handing the workflow to support."}]},{"term":"operational workflow","term_normalized":"operational workflow","senses":[{"definition":"The actual sequence of work users follow to make decisions, handle exceptions, and complete tasks — often different from what's documented.","usage":"The operational workflow included a Slack approval that was not documented anywhere."}]},{"term":"operator","term_normalized":"operator","senses":[{"definition":"A business operator: a person (or agent) who performs significant aspects of running a business. In this sense, operating a business is distinct from just founding it (creation/vision), funding it (investing money), or just working at it (as an employee).","usage":"Josephine has grown that robot laundromat business from 2 struggling locations to 10 profitable ones, so she is a real operator."},{"definition":"A technical operator. A person who interfaces with a business to figure out what needs to be automated how, then creates that automation (typically using some custom software and some existing software) and sees it through to successful production that improves the business' efficiency, profitability, or capabilities. This may be an engineer with business and communication skills and initiative.","usage":"That org moved its rote work from office tools to agents, so its people could focus on the things they do best as humans, as proof that the operator who made it happen really succeeded."},{"definition":"A sales operator. Any person in the spectrum of business growth role (e.g., from business development associate to president) who does not deliver client success directly, but instead is focused on growing the business of their firm (e.g., by finding and nurturing relationships with people whose needs they understand how to help).  In this sense, the role is technically non-technical.","usage":"Alex is such an operator, they exceeded their sales growth goal for the sixth quarter in a row."},{"definition":"Generally, an agent (human or machine) that \"runs\" (monitors and delivers using) any other thing or system.","usage":"Some people ran printed presses in the 1900s so people could get their newspapers, and some software agents ran slide deck creation in the 2020s so people could get their presentation material; both kinds entities were an operator in their own way."}]},{"term":"operator workflow","term_normalized":"operator workflow","senses":[{"definition":"The day-to-day workflow followed by frontline operators, analysts, reviewers, or support staff who will actually use the deployed system.","usage":"The operator workflow had to surface exceptions first because that is how the team worked the queue."}]},{"term":"override signal","term_normalized":"override signal","senses":[{"definition":"A signal generated when users correct, reject, or bypass system output — one of the most valuable inputs for improving evals and workflow design.","usage":"The override signal revealed that the model misunderstood one policy clause."}]},{"term":"paper cut","term_normalized":"paper cut","senses":[{"definition":"A small UX, data, or workflow friction that seems minor but harms adoption when repeated on every case.","usage":"The paper cut was one extra click on every case."}]},{"term":"partner FDE pool","term_normalized":"partner fde pool","senses":[{"definition":"A bench of partner or vendor-trained FDEs that can be assigned to customer implementations when demand exceeds the core field team.","usage":"The partner FDE pool handled connector setup while the core FDEs focused on the agent workflow."}]},{"term":"partner implementation","term_normalized":"partner implementation","senses":[{"definition":"A customer implementation delivered by a partner — often with vendor FDE guidance, playbook access, and escalation support for issues the partner can't resolve.","usage":"The partner implementation reused the vendor's deployment playbook."}]},{"term":"partner-embedded FDE","term_normalized":"partner-embedded fde","senses":[{"definition":"An FDE working through or alongside a services partner — usually to help the partner deliver production-grade implementations using the vendor's platform.","usage":"The partner-embedded FDE reviewed the partner's MCP server before the customer demo."}]},{"term":"permission model","term_normalized":"permission model","senses":[{"definition":"The access-control design that determines what users, agents, tools, and services can see or do in the deployment. Needs to be reviewed before enabling any write actions.","usage":"The FDE reviewed the permission model before enabling account updates."}]},{"term":"permissions boundary","term_normalized":"permissions boundary","senses":[{"definition":"The line an agent, service, or user cannot cross — enforced through access controls, tool scopes, policy, or runtime checks.","usage":"The permissions boundary kept the agent from reading HR data."}]},{"term":"PII handling","term_normalized":"pii handling","senses":[{"definition":"The way a deployment detects, protects, minimizes, logs, or avoids personally identifiable information — must be approved before real customer data enters any eval set.","usage":"PII handling had to be approved before real customer tickets entered the eval set."}]},{"term":"pilot-to-production path","term_normalized":"pilot-to-production path","senses":[{"definition":"The concrete sequence of work needed to move from prototype or pilot to production: owners, security, data, integration, evals, training, and support. Should be written down before the pilot starts.","usage":"The pilot-to-production path listed SSO, audit logging, and supervisor training as launch gates."}]},{"term":"platform capability","term_normalized":"platform capability","senses":[{"definition":"A reusable product or platform function that can be configured or extended for customer needs without becoming bespoke code. FDEs push custom work toward this.","usage":"The FDE reused a platform capability for approvals instead of building a one-off queue."}]},{"term":"policy check","term_normalized":"policy check","senses":[{"definition":"A validation step that confirms output, action, or data access complies with customer policy or product rules before the workflow continues.","usage":"The policy check blocked responses that lacked required disclosures."}]},{"term":"private deployment","term_normalized":"private deployment","senses":[{"definition":"A deployment isolated to a specific customer or environment for security, compliance, networking, or data-control reasons.","usage":"The bank required a private deployment before testing with production data."}]},{"term":"product gap","term_normalized":"product gap","senses":[{"definition":"A missing product capability that forces field teams into custom work or blocks customer value. Should be logged and escalated — not silently worked around every time.","usage":"The FDE logged a product gap for per-region approval policies."}]},{"term":"product-led deployment","term_normalized":"product-led deployment","senses":[{"definition":"A deployment motion where the product is self-serve or template-driven enough that less field engineering is required per customer. What FDE teams push toward after proving a pattern.","usage":"The FDE pushed the pattern toward product-led deployment after the third repeat."}]},{"term":"production credentials","term_normalized":"production credentials","senses":[{"definition":"Credentials scoped to live systems — managed, rotated, and audited more carefully than test credentials. The FDE waits for these before final validation.","usage":"The FDE waited for production credentials before final validation."}]},{"term":"production-grade agentic workflow","term_normalized":"production-grade agentic workflow","senses":[{"definition":"An agentic workflow built for real users and real consequences: evals, guardrails, monitoring, permissions, and operational ownership all in place. Typically has fewer tools than the demo.","usage":"The production-grade agentic workflow had fewer tools than the demo but was much safer."}]},{"term":"productionized agent","term_normalized":"productionized agent","senses":[{"definition":"An agent that is evaluated, observable, permissioned, supportable, and integrated into a real workflow with clear human handoffs — not just a prompt that works most of the time.","usage":"The productionized agent opened tickets only after passing policy checks."}]},{"term":"productionized AI application","term_normalized":"productionized ai application","senses":[{"definition":"An AI application hardened for real use — governed data access, evals, monitoring, support, permissions, and a clear workflow owner. Not the thing that impressed in the demo.","usage":"The productionized AI application replaced the one-off prompt chain."}]},{"term":"productization tradeoff","term_normalized":"productization tradeoff","senses":[{"definition":"The decision between solving one customer's problem quickly and building a reusable capability that scales to the next ten accounts. FDEs face this on every custom build.","usage":"The productization tradeoff was whether to hardcode the rules or build an admin UI."}]},{"term":"prompt pack","term_normalized":"prompt pack","senses":[{"definition":"A versioned set of prompts, instructions, examples, and tool guidance used by an AI application or agent. Versioned alongside the release it belongs to.","usage":"The prompt pack changed with the workflow, so the FDE versioned it with the release."}]},{"term":"proof-of-value engagement","term_normalized":"proof-of-value engagement","senses":[{"definition":"A short engagement designed to prove value on realistic customer data and workflows — not just show that the product can demo well. Ends with a launch plan, not a prettier prototype.","usage":"The proof-of-value engagement ended with a launch plan, not a deck."}]},{"term":"rapid prototype handoff","term_normalized":"rapid prototype handoff","senses":[{"definition":"The transfer of a prototype to the team that will productionize or operate it — with assumptions, known gaps, code, data dependencies, and next steps documented. Fails when the mock data source is undocumented.","usage":"The rapid prototype handoff failed because nobody documented the mock data source."}]},{"term":"read-only pilot","term_normalized":"read-only pilot","senses":[{"definition":"A pilot where the system can retrieve, reason, and recommend but cannot modify any customer systems. Lets users build trust before the agent takes actions.","usage":"The read-only pilot let users build trust before the agent could take actions."}]},{"term":"red team scenario","term_normalized":"red team scenario","senses":[{"definition":"A deliberately adversarial or high-risk test case used to probe model, agent, security, or workflow weaknesses before production.","usage":"The red team scenario checked whether the agent would reveal another account's data."}]},{"term":"reference architecture","term_normalized":"reference architecture","senses":[{"definition":"A reusable architecture for a class of deployments: system boundaries, integrations, security patterns, and common variants. Built from a real deployment, not invented from scratch.","usage":"The FDE turned the first claims deployment into a reference architecture."}]},{"term":"reference customer","term_normalized":"reference customer","senses":[{"definition":"A customer deployment strong enough to be used as evidence for other customers, prospects, or internal investment decisions.","usage":"The reference customer proved the agent could work in regulated operations."}]},{"term":"reference implementation","term_normalized":"reference implementation","senses":[{"definition":"A working implementation meant to demonstrate the recommended pattern and give future deployments a tested starting point.","usage":"The FDE wrote a reference implementation for the MCP connector."}]},{"term":"reinvention deployed engineer","term_normalized":"reinvention deployed engineer","senses":[{"definition":"A services role associated with Accenture and Anthropic's enterprise AI deployment partnership, focused on turning AI capabilities into deployed customer workflows at scale.","usage":"The reinvention deployed engineer worked with the customer team to turn a transformation idea into a deployed workflow."}]},{"term":"release train","term_normalized":"release train","senses":[{"definition":"A scheduled release process that deployments must align with in enterprise environments. Means the FDE has one launch window per month, so readiness must be real, not optimistic.","usage":"The customer release train meant the FDE had one launch window per month."}]},{"term":"reusable integration pattern","term_normalized":"reusable integration pattern","senses":[{"definition":"A repeatable approach for connecting a class of systems — auth, data mapping, error handling, operational checks — documented after the second time it's needed.","usage":"The reusable integration pattern covered every customer using the same claims platform."}]},{"term":"risk review","term_normalized":"risk review","senses":[{"definition":"A review of deployment risks — data exposure, bad actions, reliability, compliance, user misuse, and operational failure modes — often what changes a full write-back launch into an approval-only one.","usage":"The risk review changed the launch from full write-back to approval-only."}]},{"term":"roadmap feedback loop","term_normalized":"roadmap feedback loop","senses":[{"definition":"The process of turning repeated field evidence into product requirements and roadmap priorities — requires the FDE to pattern-match across accounts.","usage":"The roadmap feedback loop converted one-off fixes into a connector roadmap."}]},{"term":"ROI model","term_normalized":"roi model","senses":[{"definition":"A simple model tying deployment outcomes to business value — time saved, risk reduced, throughput, cost, revenue, or quality. Used to justify expansion, not to impress stakeholders.","usage":"The ROI model justified expanding the workflow to the second region."}]},{"term":"rollback plan","term_normalized":"rollback plan","senses":[{"definition":"A predefined way to safely revert, disable, or bypass a deployed change if it causes problems in production. Required before any write-capable agent goes live.","usage":"The FDE refused to launch the write-back action without a rollback plan."}]},{"term":"runbook","term_normalized":"runbook","senses":[{"definition":"A practical guide for operating, troubleshooting, escalating, and recovering a deployed system — written so the support team can use it without the FDE in the room.","usage":"The runbook told support how to disable write-back if the CRM API failed."}]},{"term":"sample code","term_normalized":"sample code","senses":[{"definition":"Small, working code showing the right integration pattern or API usage — not a production implementation, but enough for the customer's engineers to build from.","usage":"The sample code showed the customer how to call the approval endpoint."}]},{"term":"sandbox-to-prod gap","term_normalized":"sandbox-to-prod gap","senses":[{"definition":"The specific differences between sandbox and production that can break a deployment at launch — usually firewall rules, stricter roles, and real data shapes.","usage":"The sandbox-to-prod gap was a missing firewall rule and stricter production roles."}]},{"term":"scope-speed-scalability tradeoff","term_normalized":"scope-speed-scalability tradeoff","senses":[{"definition":"The constant FDE tension between solving the immediate customer need fast, keeping scope narrow, and building something reusable for the next account.","usage":"The scope-speed-scalability tradeoff pushed the team to hardcode one rule for launch and productize it later."}]},{"term":"scoping spike","term_normalized":"scoping spike","senses":[{"definition":"A short technical investigation used to reduce uncertainty before committing to a build plan or timeline.","usage":"The scoping spike proved the customer's API could support the workflow."}]},{"term":"semantic layer","term_normalized":"semantic layer","senses":[{"definition":"A layer that gives data business meaning through metrics, entities, relationships, definitions, and governed access — prevents the agent from guessing what terms mean.","usage":"The semantic layer stopped the agent from guessing what 'active account' meant."}]},{"term":"service account","term_normalized":"service account","senses":[{"definition":"A non-human account used by a service, connector, or agent to access customer systems under controlled permissions. Should follow least privilege.","usage":"The FDE created a service account with read-only access for the pilot."}]},{"term":"services motion","term_normalized":"services motion","senses":[{"definition":"A go-to-market and delivery motion where hands-on services, implementation, or field engineering are part of the value proposition — not an afterthought or a professional services add-on.","usage":"The services motion helped customers adopt the AI product faster than self-serve alone."}]},{"term":"shadow mode","term_normalized":"shadow mode","senses":[{"definition":"Running a new workflow, model, or agent in parallel with the current process without letting it take production actions. Used to compare outputs against human decisions before trusting the system with real consequences.","usage":"The agent ran in shadow mode for two weeks so the FDE could compare its recommendations to human decisions."}]},{"term":"solution engineer","term_normalized":"solution engineer","senses":[{"definition":"A technical customer-facing engineer who maps the customer's problem to a product solution, supporting demos, architecture, proof of value, and deployment planning. Typically pre-sale or early deployment.","usage":"The solution engineer helped qualify the use case before the FDE committed to building anything."}]},{"term":"source-of-truth dispute","term_normalized":"source-of-truth dispute","senses":[{"definition":"A disagreement about which system or dataset is authoritative for a field or decision. Must be resolved before write-back — otherwise two systems diverge and nobody owns the reconciliation.","usage":"The source-of-truth dispute had to be resolved before write-back."}]},{"term":"SSO integration","term_normalized":"sso integration","senses":[{"definition":"Connecting the deployment to the customer's single sign-on provider so users authenticate through enterprise identity controls. Usually required before the customer adds frontline users.","usage":"SSO integration was required before the customer would add frontline users."}]},{"term":"staged rollout","term_normalized":"staged rollout","senses":[{"definition":"A rollout that expands gradually across users, teams, regions, features, or permissions to control risk and learn from early usage before going broad.","usage":"The staged rollout kept write access limited to supervisors for the first week."}]},{"term":"sub-agent","term_normalized":"sub-agent","senses":[{"definition":"A smaller specialized agent or model role used inside a larger agent workflow to handle a specific task.","usage":"The sub-agent handled policy lookup while the main agent drafted the answer."}]},{"term":"success metric","term_normalized":"success metric","senses":[{"definition":"A concrete measure used to judge whether the deployment is working: hours saved, cycle time, adoption, error reduction, or revenue impact. Agreed before launch, not invented after.","usage":"The success metric was time from ticket creation to first useful response."}]},{"term":"support handoff","term_normalized":"support handoff","senses":[{"definition":"The transfer of knowledge and responsibility from the build team to whoever supports the deployment after launch — with common errors, dashboards, rollback steps, and escalation paths documented.","usage":"The support handoff included common errors, dashboards, and rollback steps."}]},{"term":"system of action","term_normalized":"system of action","senses":[{"definition":"A system where users or agents take operational actions — not just view information. The goal of most FDE deployments.","usage":"The claims platform became the system of action for approvals."}]},{"term":"system of engagement","term_normalized":"system of engagement","senses":[{"definition":"The interface where users interact with workflows, recommendations, and collaboration around work — the best place to embed an AI capability because users already live there.","usage":"The FDE embedded the assistant in the system of engagement users already opened daily."}]},{"term":"system of record","term_normalized":"system of record","senses":[{"definition":"The authoritative system for a type of business data. Matters for write-back: the FDE needs to know which system wins when there's a conflict.","usage":"The CRM was the system of record for account status."}]},{"term":"system owner","term_normalized":"system owner","senses":[{"definition":"The person or team accountable for a source system, integration, or operational dependency. Usually the one who has to approve new API scopes.","usage":"The system owner had to approve the new API scope."}]},{"term":"systems-of-action deployment","term_normalized":"systems-of-action deployment","senses":[{"definition":"A deployment that doesn't just surface information — it triggers, updates, approves, or coordinates actions in operational systems. Requires stronger permissions, audit, and fallback design.","usage":"The systems-of-action deployment wrote approved updates back into the CRM."}]},{"term":"technical artifact","term_normalized":"technical artifact","senses":[{"definition":"A concrete deliverable — diagram, runbook, eval set, integration spec, launch checklist, code sample — that helps a deployment move forward or be handed off.","usage":"The architecture diagram became the technical artifact security needed."}]},{"term":"technical success engineer","term_normalized":"technical success engineer","senses":[{"definition":"A post-sale technical owner who keeps a deployment healthy, useful, and expanding — handling adoption issues, configuration, and customer-side coordination after the initial FDE builds the thing.","usage":"The technical success engineer noticed usage dropping before the sponsor did."}]},{"term":"thin slice","term_normalized":"thin slice","senses":[{"definition":"The smallest end-to-end version of a workflow that proves the path from user input to useful output or action. Ship this before building the full thing.","usage":"The FDE shipped a thin slice before expanding to every claim type."}]},{"term":"time-to-value","term_normalized":"time-to-value","senses":[{"definition":"The time between starting the engagement and the customer seeing a useful, measurable result. FDEs optimize for this by launching one workflow instead of boiling the ocean.","usage":"The FDE optimized for time-to-value by launching one workflow instead of boiling the ocean."}]},{"term":"tool contract","term_normalized":"tool contract","senses":[{"definition":"The agreed inputs, outputs, permissions, errors, and side effects of a tool an agent or workflow can call. Makes clear what the agent can do, not just what it can call.","usage":"The tool contract made it clear the agent could create drafts but not send them."}]},{"term":"tool registry","term_normalized":"tool registry","senses":[{"definition":"A governed inventory of tools available to agents or applications — ownership, scopes, descriptions, and approval status tracked so nothing undocumented gets used in production.","usage":"The FDE removed an unsafe write tool from the tool registry."}]},{"term":"tool sandbox","term_normalized":"tool sandbox","senses":[{"definition":"A safe environment where agent tool calls can be tested without touching production data or committing real actions.","usage":"The tool sandbox let the team test refunds without issuing refunds."}]},{"term":"tool-use trace","term_normalized":"tool-use trace","senses":[{"definition":"A record of which tools an agent called, with inputs, outputs, timing, and errors — the primary debugging artifact for agentic workflows.","usage":"The tool-use trace explained why the agent chose the wrong escalation path."}]},{"term":"trusted technical advisor","term_normalized":"trusted technical advisor","senses":[{"definition":"A customer-facing technical partner who has earned enough trust to challenge the customer's assumptions, explain tradeoffs honestly, and guide the deployment toward production rather than toward a good demo.","usage":"The FDE acted as a trusted technical advisor when the sponsor wanted to skip evals."}]},{"term":"use-case development","term_normalized":"use-case development","senses":[{"definition":"Turning a raw customer idea into a scoped deployment: named users, workflow steps, data sources, actions, risks, success metrics, and a first release plan.","usage":"The FDE used use-case development to narrow five AI ideas down to one production path."}]},{"term":"user feedback loop","term_normalized":"user feedback loop","senses":[{"definition":"A mechanism for collecting user reactions, corrections, overrides, and requests and feeding them back into deployment or product improvements.","usage":"The user feedback loop showed operators wanted shorter explanations."}]},{"term":"VPC integration","term_normalized":"vpc integration","senses":[{"definition":"Connecting a deployment to customer or cloud network resources through a controlled virtual private cloud — typically needed when the agent must reach internal APIs not exposed to the internet.","usage":"VPC integration was needed so the agent could reach the internal API."}]},{"term":"white-glove deployment","term_normalized":"white-glove deployment","senses":[{"definition":"A high-touch deployment where the field team provides extra hands-on support, customization, and coordination — usually for a strategic or complex customer.","usage":"The white-glove deployment included onsite workflow mapping and daily launch support."}]},{"term":"workflow context","term_normalized":"workflow context","senses":[{"definition":"The information needed to make a workflow step useful: user role, current task, relevant object, policy, history, and next action. FDEs pass exactly what's needed, not the whole ticket.","usage":"The FDE passed workflow context into the agent instead of dumping the whole ticket."}]},{"term":"workflow KPI","term_normalized":"workflow kpi","senses":[{"definition":"A metric tied to the performance of a specific workflow — cycle time, backlog, error rate, throughput, escalation rate. Should move when the deployment works.","usage":"The workflow KPI improved after the agent prefilled the review form."}]},{"term":"workflow owner","term_normalized":"workflow owner","senses":[{"definition":"The person accountable for how a workflow runs in production: decisions, exceptions, user adoption, and success metrics. An FDE should not launch without one.","usage":"The FDE would not launch until there was a named workflow owner for escalations."}]},{"term":"workflow rollout","term_normalized":"workflow rollout","senses":[{"definition":"The staged release of a new or changed workflow to users, teams, regions, or business units.","usage":"The workflow rollout started with one region before adding the national team."}]},{"term":"workflow SLA","term_normalized":"workflow sla","senses":[{"definition":"An expected service level for a workflow — response time, turnaround time, uptime, or escalation time. Defines the reliability bar the deployment must meet.","usage":"The workflow SLA required the agent queue to fail over to manual review."}]},{"term":"write action","term_normalized":"write action","senses":[{"definition":"An operation that changes data or state in a customer system — requires stricter permissions, validation, and audit logging than read actions. Always treated with extra scrutiny.","usage":"The write action required a separate security review."}]},{"term":"write-back","term_normalized":"write-back","senses":[{"definition":"Writing an approved output, decision, or update from the deployed application back into a customer system of record. Usually gated behind human approval.","usage":"The FDE gated write-back behind supervisor approval."}]}]}