Internal Comms & Meetings

How to write a launch memo with AI

Internal Comms & Meetings3 AI tools6 steps6 friction points

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

A launch memo is the internal document that aligns your team before a product ships, a feature goes live, or a major initiative kicks off. It captures the why, what, who, and when in one place — so engineers, support, sales, and leadership are working from the same facts. Most operators write one for every significant release, which means it's a recurring task that compounds in importance as the pace of shipping increases.

The workflow feels like a natural fit for AI because it's fundamentally a structured writing task with clear inputs. You already know the launch details — the feature name, the target user, the rollout plan, the risks. The hard part is turning a brain dump of Slack threads, Notion docs, and meeting notes into a crisp, well-organized document that a busy team member can read in three minutes. That translation work is exactly what LLMs are good at.

ChatGPT, Claude, and Gemini can all produce a solid launch memo draft from a detailed prompt. They handle structure well — executive summary, background, goals, rollout plan, risks, escalation path. The output is usually coherent and well-formatted on the first pass. Where they fall short is assembly: getting the right raw material into the prompt requires manual gathering, and nothing about the process carries over to the next launch.

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

ClaudeChatGPTGemini

Step-by-step

1 Before opening any AI tool, pull together your source material manually: the Slack threads where the feature was debated, the Notion doc with specs, the Jira tickets tracking open issues, and any dates from your calendar or project tracker. Paste all of this into a single scratch doc.

2 Open Claude or ChatGPT and paste your compiled notes. Start with a framing prompt that names the audience, the stakes, and the format you want — a launch memo with specific sections, not a generic summary.

3 Review the first draft section by section. The structure will usually be right; the details will sometimes be wrong — LLMs occasionally infer facts that aren't in your notes. Correct errors and mark the sections that need additional context.

4 Paste the corrected draft back into the chat and prompt the model to sharpen specific sections: tighten the executive summary, make the rollout steps more precise, or rewrite the risk section to name actual mitigations rather than generic caveats.

5 Export the cleaned draft to Google Docs or Notion. Manually add links to the relevant tickets, specs, and dashboards the model couldn't access. Share the doc and collect comments.

6 Repeat this entire process for the next launch — the AI has no memory of the last memo, so you're reconstructing context and format instructions from scratch each time.

Prompts you can copy

Write a launch memo for our new CSV export feature. Audience: internal team of 12. Sections: executive summary, background, what's shipping, rollout plan (staged, 10% then 100%), risks, and escalation contacts. Use this raw context: [paste notes]

The risk section of this launch memo is too vague — it lists 'potential data issues' without saying who owns the fix or what the trigger is to roll back. Rewrite it with concrete owners and a clear rollback condition.

Summarize this Slack thread into 3 bullet points suitable for the 'background' section of a launch memo. Focus on the core decision and the tradeoffs we accepted: [paste thread]

Rewrite the executive summary of this memo so a non-technical stakeholder understands what's shipping, why it matters to customers, and what they need to do before launch day. Keep it under 100 words.

Based on this feature spec and these open Jira tickets, list the top 3 launch risks and suggest a mitigation or owner for each: [paste spec and ticket summaries]

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 live connection to your project tracker or Notion — you manually copy-paste ticket statuses, spec docs, and Slack threads into the prompt window before every memo.

Context assembly is the bottleneck. Finding the right Slack threads, the right spec version, and the right open tickets can take 20-30 minutes before the AI does anything useful.

The model has no memory of your company's memo format, voice, or terminology. You re-explain section structure, naming conventions, and audience context every single time.

Nothing persists. The polished memo from last quarter's launch lives in Docs; the prompt you used to write it is gone. The next launch starts from a blank chat window.

Draft quality drops when your notes are sparse or contradictory. If the Slack thread contains unresolved debate, the model often picks a side confidently rather than flagging the open question.

Distribution is fully manual. After writing the memo you're back to copying it into email, Notion, or Slack — the AI played no role in getting it to the people who need it.

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 connects to your live business data and builds software that runs continuously, so instead of re-prompting a chat window before every launch, the memo-writing workflow lives as a persistent app connected to the tools your team already uses.

Connect Notion from Starch's integration catalog and link your Google Calendar as a scheduled-sync provider — an agent can pull the relevant spec doc and the launch date automatically, without you hunting for context before each draft.

Use the Knowledge Management starter app to store your memo format, section definitions, and past examples in one searchable place. The agent references your actual standards instead of reinventing structure from a cold prompt every time.

Describe the workflow in plain English and Starch builds it: 'When a launch date is added to our project tracker, pull the linked spec, draft a launch memo using our standard format, and post a draft to Notion for review.' That automation runs on its own.

Connect your project management tool from Starch's integration catalog — the agent queries open tickets and their statuses live when drafting the risks section, so the memo reflects what's actually unresolved, not what you remembered to paste in.

The Presentation Agent (currently in development — request beta access) will turn the finished memo into a slide summary for the broader team standup without a separate formatting step.

Once the memo is approved, connect Gmail or Slack as a scheduled-sync provider and have Starch distribute it automatically to the right channels — the distribution step is part of the same workflow, not a separate manual task.

Get closed-beta access →

Toolkit