| name | slack-thread-to-crm |
| description | Use when asked to take people from a Slack thread/channel (e.g. "who's going to X", "who signed up for the course/event") and record them in the AI Shipping Labs CRM — resolve Slack users to emails, confirm platform users, add member notes + tags, and optionally hand off to plan generation. Triggered by screenshots of Slack threads or requests like "add these people to the CRM". |
| metadata | {"short-description":"Read a Slack thread, map participants to platform users, write CRM notes + tags"} |
Slack thread -> CRM
Turn "who is doing X" in a Slack thread into CRM records (member notes + tags) on the AI Shipping Labs platform. This is the composed recipe; it delegates the mechanics:
- Slack reads (find channel, read thread, resolve handles to emails):
ai-shipping-labs-slack.
- CRM reads/writes (user lookup, member notes, tags):
ai-shipping-labs-users.
- Downstream plan generation:
ai-shipping-labs-plan-import.
The user usually attaches a phone screenshot of the thread. The screenshot shows the topic and reply count but NOT every reply — always read the full thread from Slack, never work from the screenshot alone.
Workflow
1. Read the full thread
Find the channel and pull every reply. See ai-shipping-labs-slack for the commands (conversations.list, conversations.history, conversations.replies). #community = C0AFZSRAYQ4. reply_count includes the parent.
2. Classify intent (this is the judgment, do it yourself)
Read every reply and classify each person — do not regex "I'm in". Distinguish:
- committed — "yes, I am doing that", "I'll be doing it", "I'm in".
- tentative — "I think I might join, no concrete plan yet". Flag separately; do NOT fold into the committed group.
- advanced / doesn't need it — someone who already did a prior course. The host often replies in-thread that "the only new thing for you is X". Still record them, but mark separately.
Exclude the thread author / host (the person asking "who's going?").
3. Resolve each person to an email
Use ai-shipping-labs-slack — users.info by ID, or page users.list and match name / real_name / display_name.
4. Confirm each is a platform user
GET /api/users/{email} (see ai-shipping-labs-users). Existing tags are signal: ai-buildcamp, ai-buildcamp-1/2/3 => already took a prior course (corroborates "advanced"). A 404 means no platform account — surface it, don't invent one.
5. Write CRM: member note + tag
See ai-shipping-labs-users for the exact endpoints and the verify gotcha. Per person:
- Member note (
POST /api/member-notes, kind:general, visibility:internal) — always cite the source thread and quote the person's own words.
- Cohort tag (
POST /api/users/{email}/tags) — committed people get the course-cohort tag (e.g. llm-zoomcamp-2026); tentative people get the -interested variant so the committed group filters cleanly.
6. Hand off to plans (optional / usually a separate confirmed step)
"Generate plans so they can attend" is downstream. Plans attach to a sprint (GET /api/sprints). The LLM-Zoomcamp-style community sprint may not exist yet — do NOT auto-create a sprint and bulk plans without confirming the target sprint slug. Once confirmed, use ai-shipping-labs-plan-import.
Output to the user
Give a clean committed list (name + email), call out advanced/tentative people separately, confirm notes+tags landed, and name the next decision (which sprint to generate plans against). Don't silently fold tentative or advanced members into the committed cohort.
