Workflow Library

Prompts that turn your meetings into actual work done.

ONE-OFFPre-call prep
Quick pre-call brief
Before a meeting, ask Claude what's been said with this person or company so you walk in with full context.
The simplest, highest-impact use of the connector. No setup, no skill — just ask before each meeting. Claude pulls past transcripts and notes for the same attendees and gives you a one-screen brief.
I have a meeting with [name/company] at [time]. Pull our past calls together and brief me: what we've discussed, what was promised, where things stand, what to lead with.
Copy
ONE-OFFPost-call admin
Follow-up email draft
Right after a call ends, get a draft follow-up that captures the key moments and any commitments you made.
The bread-and-butter post-call workflow. Faster than rewatching the meeting, more accurate than memory, and Claude can match your voice if you give it a few examples.
Draft a follow-up email for my last meeting. Capture the key points, next steps I committed to, and anything they asked for. Keep it warm but tight.
Copy
ONE-OFFTeam alignment
Find the moment, share the clip
Remember someone said something important weeks ago? Have Claude find it, clip it, and send it.
When you need a specific decision or quote to settle a debate, Claude can search transcripts, create a clip on the actual moment, and post it to Slack — all in one chat.
Find the call about three months ago where we discussed [topic]. Pull the exact moment someone said [phrase or summary], clip it, and draft a Slack message to [person] about it.
Copy
ONE-OFFKnowledge & memory
Cross-meeting question
Ask a question that spans dozens of meetings. “Across my last 100 calls, what am I learning about our ICP?”
This is where the connector starts to feel different from a regular notetaker. Instead of searching one meeting at a time, Claude synthesizes across your whole library — themes, patterns, missing context.
Across my last 100 meetings, what am I learning about [our ICP / a competitor / a recurring objection]? Pull patterns with example quotes.
Copy
ONE-OFFDeal & account work
Account state-of-play
A single-screen view of where an account stands — across every call, commitment, and open thread.
Especially useful before a renewal or QBR. Claude joins meeting history with deal data (if HubSpot or Salesforce is also connected) and surfaces blockers, wins, and what was promised.
Summarize the state of the [Company] account. Pull every conversation we've had, where they are in the buying cycle, what we've committed to, and what's blocking the deal.
Copy
ONE-OFFCoaching
Self-coach after a call
Ask Claude to score your last call — what worked, what to fix, where you lost the room.
Most reps don't get enough coaching. The connector pulls the AI scorecard plus the transcript and tells you specifically what to do differently next time. Honest, fast, on-demand.
Score my pitch from this morning. What did I do well, what did I miss, and where did I lose them? Be honest.
Copy
SKILLPre-call prep
Daily meeting prep skill
A saved skill that briefs you on every meeting you have that day with external attendees.
Same idea as the one-off pre-call brief, saved as a reusable skill so you don't re-type it every morning. Pulls today's calendar, finds the last few recordings with the same people, and writes one brief per meeting.
Build me a daily meeting prep skill. For each of today's calendar events with external attendees: search Grain for the past three recordings with the same people, write a structured brief (themes, action items, follow-ups to raise, prep notes), and post the full brief to my Slack DMs.
Copy
SKILLPost-call admin
Custom call-summary format
Tune the post-call summary to your team's exact format — “Learnings + action items,” Slack-ready.
The default summary is good but generic. A skill lets you specify exactly how you want it formatted — what sections, what tone, where it lands. Built once, used forever.
Build me a skill that takes a meeting transcript and produces a Slack-friendly summary with two sections: 'Learnings' (3-5 bullets, what we heard from the customer) and 'Action Items' (who, what, by when). Keep it under 200 words.
Copy
SKILLResearch & insights
Customer research synthesis
Drop 20+ customer interviews into a playlist, then have a skill turn the corpus into a research report.
Real example from a security software team: 20 prospect interviews recorded, dropped into a Grain playlist, then queried through the connector for themes and anonymized quotes. The output became a published research report — some quotes are now live on their website.
I have a playlist of customer interviews in Grain. Synthesize them into a research report: top themes (with frequency), tensions or contradictions, anonymized representative quotes per theme. Format as markdown.
More involved — the prompt is a starting point; expect to iterate on format and depth.
Copy
SKILLContent & marketing
Voice-of-customer copy mining
A skill that pulls the actual phrases customers use to describe your product — for landing pages, ads, case studies.
Your customers describe your product better than you do. This skill mines transcripts for unprompted descriptions, value statements, and pain language — the raw material for good copy.
Build a skill that mines my customer calls for: (1) phrases customers use to describe our product unprompted, (2) the specific outcomes they say we delivered, (3) pain points they came in with. Group by frequency, include speaker and meeting reference for each quote.
Copy
SKILLKnowledge & memory
Personal knowledge graph
Build a wiki from your meetings that grows over time — themes, decisions, and connections across calls.
An advanced workflow: meetings flow into a local 'raw' folder, then get processed into a wiki of themes and topics that link to each other. Claude Code then uses that wiki as context for decks, product analysis, and strategy work.
Set up the foundation: create three folders — raw, wiki, output. Build a skill that ingests new meeting transcripts from raw, processes them into themed wiki articles, and links related themes across articles. Reference wiki articles when producing anything in output.
Advanced — a multi-week build. The prompt kicks off the foundation; the real complexity is iteration and the hooks that keep it fresh.
Copy
SCHEDULEDPre-call prep
Morning meeting prep (auto)
Wake up to a Slack DM with briefs for every meeting on your calendar that day. No prompting required.
The daily meeting prep skill, scheduled. Runs every weekday at 8am and sends the full brief to your Slack DMs as a canvas. By the time you sit down, your prep is done.
Schedule: every weekday at 8am. Task: pull today's external meetings from my calendar, search Grain for past recordings with the same attendees, write a brief per meeting, post the full thing as a Slack canvas DM to me.
Copy
SCHEDULEDPost-call admin
Daily action-item digest
End-of-day digest of every commitment you made, every commitment made to you, and what's still open.
A nightly run that pulls action items from every call you attended that day, deduplicates, flags anything you haven't followed up on, and lands in your inbox or Slack. Catches what falls through the cracks.
Schedule: every weekday at 6pm. Task: pull every meeting I attended today, extract action items (mine plus others'), cross-reference with anything still open from earlier this week, post a summary to my Slack DMs flagging anything I owe.
Copy
SCHEDULEDTeam alignment
Engineering daily brief
A scheduled task that compares team commitments to delivery — what moved, what stuck, who's blocked.
Instead of daily standups, a scheduled task pulls ticket movement from Linear, cross-references it with what people said they'd do in standup recordings, and surfaces blockers and accountability flags.
Schedule: every weekday at 9am. Task: pull yesterday's ticket movement from Linear, look at what each engineer said they'd do in this week's standups, compare commitments to delivery, flag blockers and where I'm holding things up. Post as Slack canvas DM.
Copy
SCHEDULEDResearch & insights
Weekly feature-request roundup
Every Monday: a list of every feature ask, bug report, and customer pain mentioned in last week's calls.
Closes the loop between customer-facing teams and product. The scheduled task scans last week's transcripts for requests and complaints, categorizes them, and ships a digest to a #product-feedback channel.
Schedule: every Monday at 9am. Task: scan all calls from the past 7 days for feature requests, bug reports, and unprompted product complaints. Categorize by theme and frequency. Post to #product-feedback with example quotes and speaker references.
Copy
SCHEDULEDContent & marketing
Competitive intelligence digest
Weekly digest of every time a competitor came up in your calls — what customers said, unfiltered.
Scan transcripts for any mention of named competitors, filter to external speakers only, summarize the takes, and post to the competitive channel. Catches everything your team would otherwise miss.
Schedule: every Friday at 3pm. Task: scan this week's calls for mentions of [competitor A, B, C, D]. Filter to external speakers only — exclude my team. Group by competitor, pull the exact quotes, post to #competitive-intel.
Copy
TRIGGEREDDeal & account work
Auto MEDDPICC hygiene
Every time a new call is recorded, Claude extracts MEDDPICC fields and updates the Salesforce opportunity.
Real customer build: every 35 minutes, Claude pulls new transcripts, extracts MEDDPICC data, compares it to the current Salesforce opp, writes back any gaps, and DMs the AE a scorecard with top three priorities. Remembers manual overrides so it doesn't overwrite human edits.
Trigger: new Grain meeting recorded. For each new call: extract MEDDPICC fields from the transcript, compare to the current Salesforce opportunity, identify gaps, draft the update for Salesforce (don't overwrite manual edits), and DM the deal AE a scorecard with top 3 priorities for the week.
A real production workflow — requires webhook setup or polling, plus Salesforce write access. Build incrementally.
Copy
TRIGGEREDPost-call admin
Per-customer summary routing
Each call gets summarized in your custom format and routed to the right customer's Slack channel.
If you run a Slack Connect channel per major customer (and an internal mirror), this workflow works out which customer the call was about and lands the summary in the matching internal channel — not a single firehose.
Trigger: new Grain meeting recorded. For each call: identify which customer it was with, run my custom summary format (Learnings + Action Items), look up that customer's internal Slack channel, post the summary there.
Multi-step routing — easier if the customer-to-channel mapping lives somewhere queryable (Airtable, a config file).
Copy
TRIGGEREDPost-call admin
Calls to support tickets
When a customer mentions a bug or feature request, a ticket gets created automatically — with the clip attached.
Transcript ingestion plus clip creation makes this real. Trigger on meeting completion, scan for support-flag phrases, create a Linear or Jira ticket with the customer's actual words and a link to the moment in the call.
Trigger: new Grain meeting recorded. For each call: scan for bug reports or feature requests. For each one, create a Linear ticket with: the exact customer quote, a clip link to the moment, the customer name, and the meeting date.
Tune the detection prompts carefully — too loose creates noise tickets, too tight misses things.
Copy
TRIGGEREDKnowledge & memory
Live knowledge-graph ingestion
Every meeting that completes flows into your wiki, gets themed, and links into the existing graph automatically.
The most advanced version of the knowledge graph use case. Webhooks fire when meetings complete, transcripts pipe to a 'raw' folder, a processing skill compiles them into the wiki, and the graph grows on its own.
Trigger: Grain webhook on meeting completion. Pipeline: fetch transcript → drop in raw folder → run wiki-compile skill → update knowledge graph → notify me if any high-priority themes emerged.
Advanced — a real engineering build. Start with the manual version (Skills) before automating the trigger.
Copy
TRIGGEREDTeam alignment
Threaded corrections back to CRM
Claude posts a deal update to Slack, the AE replies with corrections in-thread, Claude writes them back to Salesforce.
Closes the human-in-the-loop. Instead of forcing AEs into Salesforce to correct AI suggestions, they reply to the Slack message. Claude watches the thread and updates the CRM on their behalf.
After posting a deal update to Slack, monitor the thread. If the AE replies with corrections (“actually the champion is X, not Y”), update the Salesforce opportunity to match. Confirm in-thread when done.
Pattern: the Slack thread as a low-friction correction interface for any AI-suggested data write. Powerful and underused.
Copy