Buzz makes the case for one portable Markdown source
A tool that combines team chat, AI agents, and Git hosting sounds convenient. It also makes source ownership more important, not less.
A new collaboration project called Buzz has attracted attention by putting three normally separate surfaces—team chat, AI agents, and Git hosting—under one roof. The project is associated with Jack Dorsey, whose earlier work makes any new communication product worth watching. But writers and content teams should resist turning an early product headline into a settled workflow.
The durable lesson is simpler: chat messages, agent prompts, repository notes, and public posts increasingly share the same raw material. Markdown is the practical bridge. It is readable without a special app, produces useful diffs in Git, works well as agent context, and can be converted into rich text when a destination expects formatted paste.
What changes when chat, agents, and Git converge
Most teams currently copy the same update through three systems. A project note starts in a document, gets shortened for chat, and later becomes a launch post. Each copy drifts. Dates change in one place but not another. A headline becomes stronger while the supporting evidence disappears. An AI assistant summarizes an already outdated version.
| Surface | Useful for | Main risk |
|---|---|---|
| Team chat | Fast discussion and coordination | Decisions disappear in the scroll |
| AI agent | Summaries, checks, and transformations | Stale context becomes confident output |
| Git repository | History, review, and accountability | Raw files are not presentation-ready |
| Rich-text publisher | LinkedIn, Medium, Notion, and newsletters | Copy-paste can damage formatting |
A combined product can reduce switching costs, but it cannot remove the need for a canonical source. The source should survive if Buzz changes its export format, an agent integration is replaced, or the team moves its public distribution from LinkedIn to another platform.
The five-step Markdown-first workflow
- Start with a local Markdown file. Use a specific filename such as
2026-08-02-launch-status.md. A date and subject are more useful thannotes-final-v3.md. - Separate facts, decisions, and discussion. Agents may summarize discussion, but an approved decision needs an owner and a date.
- Commit before distribution. Git gives the team a reviewable checkpoint before chat edits and platform-specific rewrites begin.
- Create derivatives, not replacements. Keep the canonical file intact. Generate a short chat update and a public post from it.
- Convert at the edge. When a destination needs rich text, open md2rich, convert in the browser, paste, and inspect the preview.
Example 1: a source file built for people and agents
A vague update forces readers—and agents—to infer what matters:
## Launch update
We made good progress this week. A few items remain, but the team is
working through them. We expect to share more soon.
A structured version exposes the information needed for chat, review, and later publishing:
# Export launch status — 2026-08-02
## Decision
Ship CSV export on August 6. Defer PDF export to the next release.
## Evidence
- 18 of 20 CSV fixtures pass
- accessibility review completed August 1
- two PDF font bugs remain open
## Owners
- CSV regression suite: Mina — due August 4
- release notes: Devon — due August 5
## Open question
Should enterprise accounts receive export audit logs at launch?
The second file can be summarized without losing the decision boundary. It also gives reviewers concrete claims to verify. If an agent changes August 6 to August 5, the Git diff makes that change visible.
Example 2: derive a public post without exposing internal notes
Do not paste the internal file into a public composer. Create an explicit derivative:
# CSV export ships August 6
You asked for a faster way to move reporting data into your own tools.
CSV export arrives August 6.
The first release includes:
- filtered dataset exports
- UTF-8 encoding
- consistent date columns
- accessible progress and error states
We moved PDF export to a later release rather than ship with unresolved
font problems. Existing dashboards are unchanged.
This derivative preserves the verified launch date and scope while removing owner names, test counts, and unresolved enterprise discussion. Convert this approved Markdown with md2rich, then paste the rich text into LinkedIn, Medium, Notion, or another supported editor.
Give agents a narrow publishing contract
An agent should not decide which internal facts are safe to publish. Put the constraints in the repository beside the source:
# Public derivative rules
- Keep approved dates and product names unchanged.
- Remove employee names and internal ticket links.
- Do not invent metrics or customer quotations.
- Mark uncertain claims with [REVIEW].
- Return a patch; do not publish directly.
- A human must approve the final Markdown.
The last two rules matter most. An agent can save editing time without becoming the final authority. A reviewable patch keeps the human decision in the same history as the text.
Why client-side conversion still matters
A unified collaboration tool may hold highly sensitive material: unreleased product details, private repositories, customer notes, and agent instructions. The final conversion step should not create another unnecessary copy. md2rich performs Markdown-to-rich-text conversion in the browser. It does not require an account, upload the draft to an md2rich server, or send the text to an external rewriting model.
That boundary is intentionally narrow. md2rich is not your source of truth, approval system, or AI agent. Git preserves the source. Your team approves the claims. md2rich handles the last formatting hop.
FAQ
What is Buzz?
Buzz is a collaboration project associated with Jack Dorsey that combines team chat, AI-agent workflows, and Git hosting. Because the product is new, treat current descriptions as an early landscape signal rather than a permanent feature contract.
Why use Markdown for team updates?
Markdown remains readable as plain text, creates useful Git diffs, works as structured context for agents, and can be adapted to chat or rich-text publishing surfaces.
Should an AI agent publish directly from the repository?
No. Let the agent produce a reviewable derivative or patch. A person should approve factual claims, confidential details, owners, dates, and the final destination before anything is published.
Does md2rich upload private team notes?
No. Conversion runs client-side in your browser. md2rich does not store the draft in an account or send it to a server-side conversion service.
Keep the source portable and the final conversion private
Approve one canonical Markdown file, create a destination-specific derivative, then convert it to rich text locally for LinkedIn, Medium, Notion, or your next publishing surface.
Try md2rich