Create a blank Copilot Studio agent
Step 1 of 30
Action
Create a blank Copilot Studio agentExpected resultA blank Copilot Studio agent is visible in the captured Draft workspace.
Manual mode · literal browser construction
Build Underwriting Support Agent manually.
No PAC CLI, YAML import, or plugin architect. Follow each action and its evidence boundary, compare only the reviewed references, and stop at Draft.
Synthetic disclosure: this is qualitative workflow evidence using packaged synthetic inputs. It is not a customer KPI or a live-system result.
Progress will remain available on this page, but it will not persist after you leave.
Found something inaccurate? Use Report an issue on that step. It opens a prefilled GitHub issue for review and does not submit automatically.
Build and verify
Name Underwriting Support Manual
Step 2 of 30
Action
Name Underwriting Support ManualUnderwriting Support Manual
Expected resultThe page header shows the recorded manual build name: Underwriting Support Manual.
Enter the reviewed global instructions
Step 3 of 30
Action
Enter the reviewed global instructions# Underwriting Support Agent — Manual Global Instructions You are a read-only commercial-insurance underwriting pilot for underwriters, risk analysts, pricing analysts, and senior underwriters. Use only the packaged knowledge and operation skills. ## Fixed synthetic snapshot - Every submission, applicant, loss, claim, document, inspection, score, tier, rate factor, premium, limit, status, and date is fictional and fixed. - Do not browse for underwriting guidelines, rates, authority limits, loss data, or applicant information. Never invent or substitute evidence. - If a rule or record is not packaged, state that it requires authoritative carrier review. ## Natural-language routing - Use `risk_evaluation` for submission priority, risk scores, and tiers. - Use `pricing_recommendation` only for illustrative packaged rating factors and loss evidence. - Use `guideline_check` for stated limits, required documents, inspections, exceptions, and missing evidence. - Use `exception_review` for senior-review preparation and decision boundaries. - Commercial submission flow (Midwest Manufacturing Inc., UW-2025-100, the default; no ID needed): evaluate the application -> `submission_review`; full risk assessment and loss history -> `risk_assessment`; premium to quote -> `pricing_recommendation`; coverage structure -> `coverage_structure`; compliance and authority check -> `compliance_check`; complete underwriting summary -> `underwriting_summary`. Premiums, structures, and summaries are recommendations; the underwriter quotes and binds. ## Regulated boundaries - Never provide legal, insurance, actuarial, pricing, or financial advice. - Never quote, bind, approve, decline, issue, modify, or promise coverage, premium, terms, or authority. - An authorized underwriter and, where applicable, actuarial, legal, compliance, and authority reviewers own every coverage decision. - This pilot has no live submission, rating, policy, approval, messaging, or record-changing connection. ## Evidence-first response contract 1. Lead with the synthetic application ID and the most material source-backed risk or exception. 2. Separate submission facts, calculated score or tier, guideline comparison, and proposed review path. 3. Cite the loss, factor, limit, document, inspection, or rule used. 4. State the missing evidence and required decision authority. 5. End substantive answers with: `Synthetic underwriting evidence only; no quote, binder, approval, decline, policy change, or coverage decision occurred. Authorized human review required.` <!-- locked-preview-anchors:start --> ## Skill routing map Route from the user's natural-language intent to the correct skill below. Do not narrate internal retrieval, tool selection, restrictions, or implementation mechanics; present only the user-facing result. - `UWS-01` uses skill `risk-evaluation`. - `UWS-02` uses skill `pricing-recommendation`. - `UWS-03` uses skill `guideline-check`. - `UWS-04` uses skill `exception-review`. - `UWS-05` uses skill `submission-review`. - `UWS-06` uses skill `risk-assessment`. - `UWS-07` uses skill `coverage-structure`. - `UWS-08` uses skill `compliance-check`. - `UWS-09` uses skill `underwriting-summary`. These skill names above are the ONLY valid skill identifiers. Never invent, guess, or reference any other skill name. If the correct skill or its knowledge cannot be loaded after one retry in the same turn, say so honestly and stop. Do not answer using values you already know from these instructions or from general knowledge -- a response with no real citation is not acceptable output. <!-- locked-preview-anchors:end -->
Expected resultThe reviewed manual/GLOBAL-INSTRUCTIONS.md policy is visible or saved without unrecorded edits.
Save the instruction policy
Step 4 of 30
Action
Save the instruction policyExpected resultThe reviewed manual/GLOBAL-INSTRUCTIONS.md policy is visible or saved without unrecorded edits.
Remove default web search
Step 5 of 30
Action
Remove default web searchExpected resultThe captured inventory no longer lists the default web-search capability.
Add knowledge: aibast_underwriting-support-controls-and-review.md
Step 6 of 30
Action
Add knowledge: aibast_underwriting-support-controls-and-review.mdExpected resultThe exact knowledge file linked for this step is attached under its complete filename. Wait for processing to finish before continuing. This step does not require the final inventory or prove the other uploads.
Add knowledge: aibast_underwriting-support-synthetic-records.md
Step 7 of 30
Action
Add knowledge: aibast_underwriting-support-synthetic-records.mdExpected resultThe exact knowledge file linked for this step is attached under its complete filename. Wait for processing to finish before continuing. This step does not require the final inventory or prove the other uploads.
Add skill: aibast_compliance-check_08
Step 8 of 30
Action
Add skill: aibast_compliance-check_08Expected resultThe linked SKILL.md appears with its frontmatter name and description. Open it and compare the complete instructions with the source file. This step does not require the final inventory or prove the other uploads.
Add skill: aibast_coverage-structure_07
Step 9 of 30
Action
Add skill: aibast_coverage-structure_07Expected resultThe linked SKILL.md appears with its frontmatter name and description. Open it and compare the complete instructions with the source file. This step does not require the final inventory or prove the other uploads.
Add skill: aibast_exception-review_04
Step 10 of 30
Action
Add skill: aibast_exception-review_04Expected resultThe linked SKILL.md appears with its frontmatter name and description. Open it and compare the complete instructions with the source file. This step does not require the final inventory or prove the other uploads.
Add skill: aibast_guideline-check_03
Step 11 of 30
Action
Add skill: aibast_guideline-check_03Expected resultThe linked SKILL.md appears with its frontmatter name and description. Open it and compare the complete instructions with the source file. This step does not require the final inventory or prove the other uploads.
Add skill: aibast_pricing-recommendation_02
Step 12 of 30
Action
Add skill: aibast_pricing-recommendation_02Expected resultThe linked SKILL.md appears with its frontmatter name and description. Open it and compare the complete instructions with the source file. This step does not require the final inventory or prove the other uploads.
Add skill: aibast_risk-assessment_06
Step 13 of 30
Action
Add skill: aibast_risk-assessment_06Expected resultThe linked SKILL.md appears with its frontmatter name and description. Open it and compare the complete instructions with the source file. This step does not require the final inventory or prove the other uploads.
Add skill: aibast_risk-evaluation_01
Step 14 of 30
Action
Add skill: aibast_risk-evaluation_01Expected resultThe linked SKILL.md appears with its frontmatter name and description. Open it and compare the complete instructions with the source file. This step does not require the final inventory or prove the other uploads.
Add skill: aibast_submission-review_05
Step 15 of 30
Action
Add skill: aibast_submission-review_05Expected resultThe linked SKILL.md appears with its frontmatter name and description. Open it and compare the complete instructions with the source file. This step does not require the final inventory or prove the other uploads.
Add skill: aibast_underwriting-summary_09
Step 16 of 30
Action
Add skill: aibast_underwriting-summary_09Expected resultThe linked SKILL.md appears with its frontmatter name and description. Open it and compare the complete instructions with the source file. This step does not require the final inventory or prove the other uploads.
Review model, skills, knowledge, and safety boundaries
Step 17 of 30
Action
Review model, skills, knowledge, and safety boundariesExpected resultThe screenshot visibly confirms Claude Opus 5, 9 rendered skills, and no tools. Knowledge is verified separately in the two knowledge-upload checkpoints.
Review the skill list (9 skills)
Step 18 of 30
Action
Review the skill list (9 skills)Expected resultThe screenshot visibly confirms Claude Opus 5, 9 rendered skills, and no tools. Knowledge is verified separately in the two knowledge-upload checkpoints.
Review the rest of the skill list
Step 19 of 30
Action
Review the rest of the skill listExpected resultThe screenshot visibly confirms Claude Opus 5, 9 rendered skills, and no tools. Knowledge is verified separately in the two knowledge-upload checkpoints.
Open a fresh Preview conversation
Step 20 of 30
Action
Open a fresh Preview conversationExpected resultA fresh Preview surface or its recorded qualitative result is visible; only the evidence file defines a pass.
Pass UWS-01: Underwriter
Step 21 of 30
Action
Pass UWS-01: UnderwriterWhich submission needs the most experienced underwriter, and why?
Expected resultThe captured Preview evidence records UWS-01 with the recorded identifiers UW-2025-103, Substandard; do not infer results beyond it.
Before this case
Open Preview in a fresh conversation before running UWS-01. A previous response must not influence the evidence.
Pass UWS-02: Pricing Analyst
Step 22 of 30
Action
Pass UWS-02: Pricing AnalystWalk me through the rating factors and loss evidence for Riverside without issuing a quote.
Expected resultThe captured Preview evidence records UWS-02 with the recorded identifiers UW-2025-101, Indicated Premium; do not infer results beyond it.
Before this case
Open Preview in a fresh conversation before running UWS-02. A previous response must not influence the evidence.
Pass UWS-03: Risk Analyst
Step 23 of 30
Action
Pass UWS-03: Risk AnalystWhich applications are outside a stated guideline or missing required evidence?
Expected resultThe captured Preview evidence records UWS-03 with the recorded identifiers UW-2025-103, High-Risk Specialty; do not infer results beyond it.
Before this case
Open Preview in a fresh conversation before running UWS-03. A previous response must not influence the evidence.
Pass UWS-04: Senior Underwriter
Step 24 of 30
Action
Pass UWS-04: Senior UnderwriterPrepare the exception file I need to review and state whether any coverage decision was made.
Expected resultThe captured Preview evidence records UWS-04 with the recorded identifiers UW-2025-103, No approval; do not infer results beyond it.
Before this case
Open Preview in a fresh conversation before running UWS-04. A previous response must not influence the evidence.
Pass UWS-05: Underwriter
Step 25 of 30
Action
Pass UWS-05: UnderwriterHelp me evaluate this commercial insurance application and tell me what is missing.
Expected resultThe captured Preview evidence records UWS-05 with the recorded identifiers Midwest Manufacturing Inc., 95% complete, 68/100 (Moderate); do not infer results beyond it.
Before this case
Open Preview in a fresh conversation before running UWS-05. A previous response must not influence the evidence.
Pass UWS-06: Risk Analyst
Step 26 of 30
Action
Pass UWS-06: Risk AnalystShow me the full risk assessment and the loss history for the Midwest Manufacturing submission.
Expected resultThe captured Preview evidence records UWS-06 with the recorded identifiers Industry hazard, $237K, 0.42 vs class 0.58; do not infer results beyond it.
Before this case
Open Preview in a fresh conversation before running UWS-06. A previous response must not influence the evidence.
Pass UWS-07: Underwriter
Step 27 of 30
Action
Pass UWS-07: UnderwriterWhat coverage structure, endorsements, and subjectivities do you recommend for Midwest Manufacturing?
Expected resultThe captured Preview evidence records UWS-07 with the recorded identifiers Equipment breakdown, $5K deductible, $8.2M total; do not infer results beyond it.
Before this case
Open Preview in a fresh conversation before running UWS-07. A previous response must not influence the evidence.
Pass UWS-08: Senior Underwriter
Step 28 of 30
Action
Pass UWS-08: Senior UnderwriterCheck compliance and authority before we finalize the Midwest Manufacturing recommendation.
Expected resultThe captured Preview evidence records UWS-08 with the recorded identifiers Within $10M limit, OH rates filed, 30 days; do not infer results beyond it.
Before this case
Open Preview in a fresh conversation before running UWS-08. A previous response must not influence the evidence.
Pass UWS-09: Senior Underwriter
Step 29 of 30
Action
Pass UWS-09: Senior UnderwriterGenerate the complete underwriting summary for the Midwest Manufacturing submission.
Expected resultThe captured Preview evidence records UWS-09 with the recorded identifiers Approve with conditions, $88,100, drafted for you to issue; do not infer results beyond it.
Before this case
Open Preview in a fresh conversation before running UWS-09. A previous response must not influence the evidence.
Confirm Draft and stop before publish
Step 30 of 30
Action
Confirm Draft and stop before publishExpected resultThe agent remains Draft and no Publish action is taken.
Troubleshooting
A screenshot or browserfilm frame is missing
Do not invent, recreate, or substitute an image. Keep the step open until its actual evidence is captured and reviewed. Remove --allow-pending only after all workshop evidence gates are met.
A knowledge file is still processing
Wait for ingestion to finish before Preview. A partial answer is not evidence.
A skill upload fails
Use the linked raw SKILL.md. Fix the reviewed source deliberately; do not silently skip the action.
The model differs from Easy mode
Record the substitution and stop the parity claim until Easy and Manual use the same reviewed model.
The Preview answer misses an identifier
Mark the recorded case failed, inspect instructions and inventory, then replay the exact prompt in a fresh conversation.
Should I publish?
No. Keep the same owned manual agent in Draft. Do not choose Publish as part of this tutorial.