The Offer Adapter: a follow-up gameplan after every recorded call

Use the Autoploy MCP for every read and save in this task. If it is not connected, stop and tell me instead of improvising. What to build: The Offer Adapter: a follow-up gameplan after every call. After every recorded call, draft a follow-up gameplan from the transcript in your saved voice and structure, delivered by email and logged in your Gameplans list, then send one nudge if it is still sitting three days later. Runs on your Claude subscription, or on your own AI key. Autoploy never bills you for the AI. This is for founders and non-technical builders. How it behaves: 1. Runs when your call recorder posts a finished recording. 2. Read the finished call: who was on it, how long it ran, and the full transcript. 3. If first time seeing this call, then go to "A real outside call?", otherwise go to "Already handled". 4. This call already has a gameplan, so nothing runs twice. 5. If A real outside call, then go to "Read your playbook", otherwise go to "Skip". 6. Internal or very short, so no gameplan is written. 7. Read the saved business fields this automation is locked to: your voice, offer, audience, method, proof, guardrails, and links, plus any past gameplans for this company. 8. Read the transcript and pull out the evidence: the exact quotes, commitments, objections, signals, and numbers. 9. Check every quote against the real transcript in plain code, drop any that do not match, and score how much real signal the call had. 10. If enough real signal for a full gameplan, then go to "Write the gameplan", otherwise go to "Write an honest short note". 11. Write a short, honest note: the one thing worth doing, or plainly why there is nothing here worth a full gameplan. 12. Write the full gameplan from the verified evidence only, in your voice and your gameplan structure: the funnel system, where the deal stands, the phased plan, and the offer. 13. Scan the whole draft in code for banned phrases, length, and house style. 14. If passes the slop gate, then go to "Save it to your Gameplans list", otherwise go to "One clean-up pass". 15. Give Claude the exact problems and ask for one corrected pass; if it still does not pass, ship it flagged for your eye. 16. Add one row to the Gameplans list, keyed by the recording, so a replayed call never writes a second gameplan. 17. Email the finished gameplan to you, once and only once per call. 18. Wait three days, then check whether the follow-up went out. 19. If follow-up marked sent, then go to "Loop closed", otherwise go to "One nudge". 20. You sent the follow-up, so nothing more is needed. 21. Email one short nudge that the gameplan is still sitting, and mark it nudged. Never a second nudge. 22. One nudge sent, and the automation never nags again. The tools: - Fathom (for "Where do your recorded calls come from?"). - Anthropic (for "What should read the call and write the gameplan?"). I will need you to create the key ANTHROPIC_API_KEY (Anthropic API key). I add the values in Autoploy myself; you only need their names. - Resend (for "Where should the gameplan land?"). I will need you to create the key RESEND_API_KEY (Resend API key), provide the setting NOTIFY_EMAIL_FROM (From address), provide the setting NOTIFY_EMAIL_TO (Your inbox). I add the values in Autoploy myself; you only need their names. Shared memory: This automation remembers its work in a list called "Gameplans" (Every follow-up gameplan The Offer Adapter has drafted, one row per recorded call.) that I can open in the Context section of my dashboard and see every row. Its columns are: who ("Who"), company ("Company"), call_time ("Call time"), call_title ("Call title"), gameplan ("Gameplan"), status ("Status"), recording ("Recording"). Before anything runs, make sure the list exists: check with list_collections over my Autoploy MCP, and create it with create_collection (exactly that name, that purpose, and those columns) only if it is missing. Never create a second list under a similar name; reuse this one. The platform never creates a list silently at run time, so a missing list stops the work with a plain refusal. In the automation's code, write rows through ctx.collections.insert with an idempotency key built from the id of the thing the row is about (for example the booking id), so a replayed trigger can never write the same row twice. Read prior rows with ctx.collections.query when the automation must know what it already did. The platform: Connect to my Autoploy MCP server, call get_authoring_guide to read the authoring guide, and build the automation with the Autoploy SDK contract. Model every yes/no fork as a condition step with truthful labeled branches, keep every external side effect idempotent and keyed, and deploy it with deploy_automation. Before you build: Repeat the plan back to me in plain English and ask follow-up questions about anything unclear or ambiguous. Do not guess. Ask me for each key by name when it is time, and never ask me to paste a key anywhere except my own secure setup.