Internal Comms & Meetings

How to run an async standup with AI

Internal Comms & Meetings3 AI tools6 steps6 friction points

Want an agent to handle this on your live data? See the Starch version →

An async standup is a structured daily check-in where each team member answers the same three questions — what did I complete, what am I working on today, what's blocking me — without scheduling a synchronous call. For small teams and distributed operators, it replaces the 15-minute morning meeting with something people fill out on their own time and a summary someone (usually the founder) reads and acts on.

The workflow feels automatable because the structure is so consistent. Same questions every day, same format, same distribution. Writing the prompts, collecting responses, synthesizing blockers, and sending the summary back to the team is exactly the kind of repetitive, text-heavy work that looks like an LLM problem — especially when the team is small and nobody wants to spend 20 minutes each morning on coordination overhead.

ChatGPT, Claude, and Gemini can genuinely help here. You can use them to draft the standup prompt your team fills out, summarize a batch of text responses you paste in, extract blockers and action items from a messy thread, or write a clean digest to share back to Slack. The models are good at all of this. The friction is in everything around the prompting — collecting responses, remembering to run it, distributing the output.

Skip the manual work → See the Starch version

Internal Comms & Meetings3 AI tools6 steps6 friction points

AI walkthrough

How to do it with AI today

A practical walkthrough using ChatGPT, Claude, and other off-the-shelf LLMs — what they're good at, what you'll have to do by hand.

Tools that work for this

ChatGPTClaudeGemini

Step-by-step

1 Open ChatGPT or Claude and draft a standup prompt template: ask it to write a daily async standup form with fields for yesterday, today, and blockers — plus any custom fields specific to your team (e.g., 'any customer escalations?').

2 Copy that prompt into whatever collection channel you already use: a Slack message you post manually each morning, a recurring Google Form, or a simple email you send to the team.

3 At the end of the day (or the following morning), copy all team responses — from Slack replies, email threads, or a Google Sheet export — into a fresh ChatGPT or Claude window.

4 Paste a summary prompt: ask the model to extract all blockers, group updates by person, flag anything that looks like a dependency or cross-team issue, and produce a clean digest paragraph.

5 Review the output, manually add any context the model missed (it only knows what you pasted), and distribute the summary — copy it into Slack, email, or Notion by hand.

6 Repeat from step 2 tomorrow. Nothing from today's session carries over; you'll re-paste the same instructions and context each time.

Prompts you can copy

Here are async standup updates from my team of 5. Extract all blockers, group by person, and write a 3-sentence summary I can post to our #general Slack channel.

Write a daily async standup template for a 6-person software team. Include fields for: completed yesterday, planned today, blockers, and any decisions needed from me as the founder.

Below are 5 standup responses. Identify any cross-team dependencies or tasks where two people are working on the same thing without coordinating. Flag them clearly.

Rewrite these raw standup responses into a clean digest email I can forward to the full team. Keep it under 200 words, use bullet points, and surface any blockers at the top.

Given these standup updates, which tasks are overdue based on what people said they'd finish yesterday versus what they completed? List them by owner.

Reality check

Where this gets hard

The walkthrough above works — until your numbers change, the LLM hallucinates, or you have to re-paste everything next month.

No connection to your project management tool — you paste in text by hand, so the model can't see what's actually assigned, overdue, or in progress in Linear, Asana, or Jira.

Collection is still manual. The LLM summarizes responses you gather; it doesn't ping your team, collect their updates, or know when everyone has replied.

Nothing persists between sessions. The summary you built on Tuesday is gone Wednesday — you're re-pasting instructions, re-establishing context, and starting over every single day.

Output format drifts. The digest structure you carefully prompted this week may not match what you get next week if you change the wording even slightly, making it hard to skim consistently.

Blockers don't go anywhere. The model surfaces them in a summary, but there's no link to a task, no assignment to a person, and no follow-up if the blocker is still there tomorrow.

Distribution is a separate manual step every time — copy the output, open Slack or email, paste it, post it. The model produces text; getting it to your team is still on you.

Tired of the friction?

Starch runs the whole workflow on live data — no copy-paste, no hallucinated numbers, no re-prompting next month.

See the Starch version →

Starch alternative

The same workflow on Starch

Starch is an agentic operating system — it builds and runs the persistent software your workflow depends on. For async standups, that means an agent builds the standup app connected to your live Slack, project data, and task tracker, runs it on a schedule, and distributes the digest automatically — without you re-prompting anything.

Connect Slack from Starch's integration catalog and Google Calendar as a scheduled-sync provider — the agent collects standup responses from a Slack channel and cross-references today's calendar to know what each person had scheduled.

Use the Project Management starter app to give the standup context: the agent can see what tasks are open, overdue, or blocked in your Starch project board when it writes the daily digest — no copy-paste required.

Describe the standup flow you want in plain English: 'Every weekday at 9am, collect standup replies from #standup in Slack, extract blockers, and post a digest to #founders-summary' — Starch builds and schedules that automation once, and it runs continuously.

Blockers extracted from the standup can automatically create tasks in Starch's Task Manager or Project Management app — assigned to the right person, with the blocker as the task description — so they don't disappear into a digest nobody revisits.

The Meeting Notes app captures any follow-up sync calls in full, then the standup automation can pull last week's unresolved action items into context when it builds the next digest — so recurring blockers actually get noticed.

Get closed-beta access →

Toolkit