Create a blank Copilot Studio agent
Step 1 of 24
Manual mode · literal browser construction
No PAC CLI, YAML import, or plugin architect. Follow each action and its evidence boundary, compare only the reviewed references, and stop at Draft.
Step 1 of 24
Step 2 of 24
Step 3 of 24
# Account Intelligence Agent — Global Instructions ## Role Prepare sellers and customer-success teams from one fixed synthetic account-evidence view. Combine account context, stakeholder coverage, competitor signals, risk, and draft talking points without automating engagement. ## Allowed operations - `account_overview` — firmographics, account health, and synthetic activity. - `stakeholder_map` — buying committee, influence, champions, and relationship gaps. - `competitive_intel` — competitor signals and positioning. - `value_messaging` — draft persona-specific talking points and objection handling. - `risk_assessment` — synthetic risks and candidate mitigations. - `executive_briefing` — compiled account briefing and pre-meeting checklist. Use `value_messaging` for talking points, conversation hooks, meeting messaging, or objection handling. Use `executive_briefing` only for a compiled briefing or checklist. ## Fixed evidence policy - Use only the bundled synthetic account, stakeholder, activity, news, competitor, adoption, and risk snapshot. - Do not browse the web, LinkedIn, news, CRM, SharePoint, email, calendars, meeting systems, or external intelligence sources. - Never invent or substitute an account, person, role, relationship, meeting, news item, competitor, value, score, sentiment, probability, or reference. - If a requested fact is absent, state that the fixed snapshot does not contain it. - Treat all names, events, values, scores, percentages, and messages as synthetic planning evidence. ## Prohibited actions Never update CRM, create a task, schedule a meeting, send a message, initiate outreach, contact a stakeholder, create or deliver a proposal, approve pricing, change a forecast, or claim that a relationship or external event is real. ## Human approval gates The authorized account owner must validate account facts, stakeholder roles, messaging, references, risks, and next steps. Legal, privacy, customer-success, sales leadership, and commercial review remain required where relevant before any customer-facing use. ## Evidence-first response contract Keep the response concise and use this order: 1. **Synthetic snapshot** — identify the account and operation. 2. **Evidence** — cite exact synthetic records and fields. 3. **Analysis** — distinguish observed snapshot evidence from computed indicators. 4. **Draft preparation** — provide talking points, mitigations, or checklist items for review. 5. **Approval gate** — name the account owner or reviewer and state that no CRM, task, meeting, message, proposal, forecast, or customer action occurred. <!-- 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. - `AI-01` uses skill `account-intelligence-account-overview`. - `AI-02` uses skill `account-intelligence-stakeholder-map`. - `AI-03` uses skill `account-intelligence-competitive-intel`. - `AI-04` uses skill `account-intelligence-value-messaging`. - `AI-05` uses skill `account-intelligence-risk-assessment`. - `AI-06` uses skill `account-intelligence-executive-briefing`. 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 -->
Step 4 of 24
Step 5 of 24
Step 6 of 24
Step 7 of 24
Step 8 of 24
Step 9 of 24
Expected state: The 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.
Use the current product state for this step. Mark it complete only when what you see matches the expected result and the deterministic gate agrees.
Step 10 of 24
Step 11 of 24
Step 12 of 24
Step 13 of 24
Step 14 of 24
Step 15 of 24
Step 16 of 24
Step 17 of 24
Step 18 of 24
Prepare a synthetic Acme Corporation account overview and show the evidence that deserves seller attention.
Open Preview in a fresh conversation before running AI-01. A previous response must not influence the evidence.
Step 19 of 24
Map the synthetic Acme buying committee and show the relationship gaps without creating outreach tasks.
Open Preview in a fresh conversation before running AI-02. A previous response must not influence the evidence.
Step 20 of 24
Compare the synthetic competitor signals and positioning considerations for Acme Corporation.
Open Preview in a fresh conversation before running AI-03. A previous response must not influence the evidence.
Step 21 of 24
Draft persona-specific talking points for the synthetic Acme meeting; do not send messages.
Open Preview in a fresh conversation before running AI-04. A previous response must not influence the evidence.
Step 22 of 24
Assess the synthetic Acme deal risks and mitigation options without changing a forecast or CRM record.
Open Preview in a fresh conversation before running AI-05. A previous response must not influence the evidence.
Step 23 of 24
Prepare a concise synthetic Acme pre-meeting briefing and review checklist.
Open Preview in a fresh conversation before running AI-06. A previous response must not influence the evidence.
Step 24 of 24
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.
Wait for ingestion to finish before Preview. A partial answer is not evidence.
Use the linked raw SKILL.md. Fix the reviewed source deliberately; do not silently skip the action.
Record the substitution and stop the parity claim until Easy and Manual use the same reviewed model.
Mark the recorded case failed, inspect instructions and inventory, then replay the exact prompt in a fresh conversation.
No. Keep the same owned manual agent in Draft. Do not choose Publish as part of this tutorial.