Meeting Notes to Asana Tasks: Make Blueprint with AI

Every meeting produces commitments, and most of them evaporate before anyone opens a task manager. This Make blueprint catches them: send a transcript in, get one Asana task per action item out — with due dates resolved and context attached — plus a Slack recap for the team.

What this scenario does

• Custom webhook receives the meeting payload: meeting_title, meeting_date, transcript
• OpenAI module extracts action items as strict JSON — for each: imperative title, assignee (as named in the meeting), due_date (ISO, with relative phrases like “next Friday” resolved against the meeting date), and 1-2 sentences of notes context
• Parse JSON module makes the array mappable
• Iterator loops over each action item
• Asana module creates a task per item, with the meeting title and date stamped into the description
• Slack module posts a one-line confirmation per task to #meeting-follow-ups

The extraction prompt is deliberately strict: it forbids invented tasks and returns an empty array when a meeting genuinely produced no commitments — so a rambling all-hands doesn’t generate task spam.

Prerequisites

• A Make account (free tier handles this comfortably at low volume)
• An OpenAI connection in Make (API key)
• Asana connection, plus your workspace ID and the project where tasks should land
• Slack connection with access to your follow-ups channel
• A transcript source: Fathom, Fireflies, tl;dv, Zoom, or Whisper output — anything that can POST to a webhook or be wired through their native Make modules

How to import

1. Download the blueprint JSON from the box at the top.
2. In Make, create a new scenario, click the three-dot menu (More) → Import Blueprint, and upload the file.
3. Re-link each module to your own connections (YOUR_OPENAI_CONNECTION, YOUR_ASANA_CONNECTION, YOUR_SLACK_CONNECTION) — Make prompts you module by module.
4. Create the webhook in module 1 and copy its URL into your transcript tool.
5. Run once with a short test transcript before scheduling.

What to customize

• Asana workspace and project IDs in module 5 — and map assignee to real Asana users if your team names are unambiguous
• The action-item definition in the system prompt — e.g., exclude “decisions” or include “open questions” as a separate type
• Slack channel and message wording
• Max tokens (1,500 default) if your meetings routinely produce 15+ action items

The step-by-step version of this build, including transcript-tool wiring, lives in the meeting notes to tasks tutorial.

Cost per run

A 45-minute meeting transcript is roughly 6,000-9,000 words, so expect 8,000-12,000 input tokens per run on gpt-4o-mini — as of mid-2026 that’s around one cent per meeting. On the Make side, each run consumes one operation per module per action item iterated, so a 6-task meeting uses about 15-20 operations; budget accordingly on the free tier’s 1,000 monthly operations.