Skip to content

Enterprise Pilot Roadmap

English | 中文

05PILOT ROADMAP

Advance through exit criteria instead of starting with a universal Agent

Add one risk class at a time: prove the task and evidence, then read-only tools, controlled writes, recovery, and multi-team composition.

Scope

One task unit, one owner, and one completion definition.

Control

Capabilities grow by stage; high-risk actions do not skip stages.

Decision

Observable evidence decides whether to continue, correct, or stop.

Stage roadmap

The roadmap does not prescribe a fixed number of weeks. System integration, data classification, and security review vary widely between enterprises. Define entry conditions, allowed capabilities, and exit evidence for each stage instead.

01
0 Frame task
02
1 Observe
03
2 Read tools
04
3 Controlled writes
05
4 Recover and scale
06
Expand or stop
The staged exit path for an enterprise pilot
StageAllowed capabilityRequired evidenceExit condition
0 Frame the taskNo production-system integrationTask samples, human baseline, completion definition, risk classificationThe team can consistently judge whether one run is complete
1 Observe onlyFixed material and model, no state-changing tool writesSession records, error types, human-rating disagreementsOutput form is stable and main failures are classifiable
2 Read-only toolsSearch, read, query, and controlled computationTool trajectory, citation or query evidence, takeover recordsFindings return to sources and errors do not change business state
3 Controlled writesReversible or low-risk writes with per-action approvalApproval pairs, business idempotency, compensation and exception recordsUnapproved calls have no effect and retries do not duplicate effects
4 Recovery and scalePersistent sessions, continuable subagents, second presetRecovery drill, capability isolation, cost and capacity observationCrash and switch behavior match expectations; a second scenario does not fork the core

Phase 0: Express the task as an acceptable unit

Collect real historical samples and record the human process, input material, allowed actions, and final artifact. Replace “improve efficiency” with an observable question such as “find all three field classes in the supplied material and provide a source for each,” rather than inventing a benefit percentage first.

This stage also defines stop conditions. If inputs are repeatedly missing, business rules have no owner, outcomes cannot be reviewed, or one task contains inseparable high-risk actions, return to process governance instead of handing all uncertainty to the model.

Phase 1: Observe the model and session only

Process samples with a fixed Agent composition but no tools that change business state. Check whether the session completely records input, model output, errors, and corrections. Establish failure classes instead of reporting only “answer satisfaction.”

At minimum, distinguish missing material, ambiguous task, model reasoning error, unmet tool need, conflicting domain rule, and output-format error. Classification identifies whether to change process, prompt, tool, or model instead of calling every issue “model instability.”

Phase 2: Connect read-only tools

Wrap the most stable sources as read-only tools: repository search, document read, ticket query, or data snapshot. Each tool needs a clear schema, timeout, errors, and presentation intent. A reviewer should be able to locate the tool result instead of receiving only a model summary.

The read-only stage validates whether answers have evidence and failures can transfer to a person. If the Agent cannot cite sources or distinguish no result from system error, writes will only amplify uncertainty.

Phase 3: Add reversible controlled writes

Prefer drafts, suggestions, temporary labels, and other reversible actions. Give each write tool policy and approval; the business interface uses caller identity, idempotency, and an explicit transaction result. Approval copy names the object and impact instead of asking only whether to “allow the tool.”

Exercise denial, unavailable approvers, tool timeout, partial business failure, and duplicate requests. Exit evidence compares live state with the log: unapproved actions have no side effect, successful calls have one business effect, and failures give an operator a clear recovery entry.

Phase 4: Validate recovery, delegation, and multi-team composition

After the main path stabilizes, validate session recovery, continuable subagents, and a second Agent preset. Recovery drills cover process termination, persistence reads, and follow-up messages. Delegated work completes under fixed authority and returns to the parent Agent when it needs more access.

The second preset should serve a different task with a different tool set. If the new scenario still copies the agent loop, session, or Host, capability seams and plugin composition have not delivered the intended foundation.

Pilot scorecard

The scorecard defines metrics and evidence sources without fabricating target values. The pilot owner sets thresholds from the human baseline, risk tier, and sample distribution.

DimensionSuggested metricEvidence source
Task qualityShare of samples accepted against the checklist; reasons for human reworkBusiness acceptance sheet and final artifacts
TraceabilityShare of key findings that return to a source or tool resultSession events, citations, and tool cards
Control effectivenessWhether denial, approval, retry, and exception branches behave as specifiedApproval events and business-system state
RecoverabilityShare of specified faults that restore context or explicitly report non-recoveryPersistence drills and recovery records
Human loadWhere approval, takeover, review, and correction time occursOperation records and interviews, not total duration alone
Run costModel tokens, tool time, remote environment, and storage consumptionRequest events, tool timing, and platform billing

Roles and responsibilities

RoleDecision it must ownResponsibility it must not transfer
Business ownerTask boundary, completion, exceptions, and final acceptanceLetting the model define business correctness
Product/process ownerHuman takeover, approval copy, and delivery interfaceHiding an undefined process in prompt wording
Platform engineeringPresets, capability providers, persistence, and observabilityInventing authorization and transaction semantics for business systems
Security and complianceData classification, execution environment, credentials, and retentionTreating session-log existence as completed compliance
Evaluation ownerSamples, human baseline, failure classes, and thresholdsUsing one average score to hide high-risk failures

Expand, correct, or stop

When should the pilot expand?

The current task meets exit criteria on the specified sample, main failures are classifiable and transferable, control branches have been exercised, and a new scenario can reuse the runtime spine.

When should the pilot correct course?

When value exists but evidence breaks, tool interfaces are unstable, approval load is too high, or one failure class dominates, repair the process, tool, or composition and repeat the stage.

When should the pilot stop?

Stop when completion remains undefined, irreversible work cannot be separated, data or identity conditions prohibit the design, takeover costs exceed task value, or value depends only on untestable future assumptions.

PILOT PRINCIPLE

Prove judgeable, controllable, and transferable before faster, broader, and more autonomous

Plugins and multi-Agent capability make scope easy to expand. Pilot discipline requires every expansion to rest on passed evidence instead of using more features to hide an unresolved earlier stage.

Evidence status: Harness capabilities and limitations are confirmed against current source. Stages, metrics, and organization advice are a pilot method; the target enterprise sets samples and thresholds.