Stop screenshotting reports into Slack

Every team does it. The dashboard updates. The Monday report runs. The eval finishes. Someone takes a screenshot, drops it into a channel, types "thoughts?" — and waits.

This is the workflow Comma was built to replace. Not because screenshots are evil, but because they're the lowest-common-denominator answer to a question that deserves a better one: how does the team see, discuss, and keep track of the work we share with each other?

What screenshots actually cost you

Each individual screenshot is fine. The pattern is what compounds:

1. Detail is lost the moment the screenshot is taken

Charts pixelate. Table rows become unsearchable. Numbers blur. Links are dead. Whoever clicks the screenshot to expand it sees a worse version of the original report.

2. Feedback fragments across thread replies

Three people see the screenshot. Each one notices a different problem. Each one leaves a reply on a different message in the thread. Now the feedback for "the report" is split across the thread tree, with no canonical place to read it back.

3. None of the feedback is anchored

When someone replies "the conversion column looks off," there is no mechanism to mark which row in the conversion table they meant. Whoever follows up has to ask. Whoever follows up after that asks again.

4. The canonical artifact stops existing

Next Monday, a new screenshot. The previous week's screenshot is now "the report two weeks ago," buried in scrollback. There is no link to "the report." There is only "the report you scrolled past on the 14th."

5. The conversation doesn't compound

Two months in, the channel is a wall of similar-looking dashboards. Nobody can answer "what did we decide about the Q3 funnel anomaly?" without a slack-search safari that lasts the rest of the afternoon.

6. Agents have nowhere to look

If your team is building agents that should respond to feedback, those agents can't read scattered Slack-thread replies. They need anchored comments on the artifact. Screenshots don't have that.

The shape of a better default

A working default for "I made a report and want my team to react to it" needs six things:

1. The original artifact, rendered. Not a screenshot — the actual HTML, displayed faithfully.
2. A stable link. Bookmarkable. The link doesn't change next week.
3. Anchored comments. Pinned to specific text and table cells, so feedback maps to specific content.
4. A canonical thread per report. All comments on one artifact, in one place.
5. Persistence across refreshes. When the report updates next week, the same link refreshes; last week's resolved threads stay in revision history.
6. Reachable by agents. If you're building agents, they need to be able to read the comments and respond.

Screenshots in Slack have zero of these. Most "let me email the file" defaults have one. A custom-built solution can have all six — and a year of maintenance work.

Comma was built so the default has all six out of the box.

How the workflow changes

The Slack message still happens. It just stops being the artifact and goes back to being a notification.

The mental model

Slack is a notification layer. It is the answer to the question "how should I find out something happened?"

Slack is not a document. It is not the answer to "where does the artifact live?"

Conflating those two questions is the root cause of the screenshot workflow. Comma separates them: Slack still notifies the team that the report is ready. The report itself — the HTML, the comment layer, the revision history — lives at a link.

This isn't a "stop using Slack" pitch. Use Slack for what Slack is good at. Use a document surface for what a document surface is good at.

What about Notion / Google Docs / a wiki?

Closer, but they can't render arbitrary HTML faithfully. Your dashboard export, your notebook render, your AI tool's output — pasting it into Notion strips the styling and the charts. You'd have to rebuild the artifact inside Notion, which costs you the time and the rendering fidelity.

Comma is the surface for the HTML you already produced.

A note on screenshotting AI tool output

This is the version of the pattern that's growing fastest. Claude or ChatGPT generates a useful HTML report. The user screenshots part of it into Slack. The conversation that follows is anchored to a static image of a dynamic artifact.

The same pattern, the same cost. And there's no read path back from "someone's slack reply" to "the part of the agent's output they were talking about." If you want your team's feedback to actually train the next run of the agent, you need anchored comments and a stable artifact — which is exactly what Comma's MCP server was built around.

See Where should my AI agent post its work? →.

Try the change

The fastest way to feel the difference is to take the next report you would have screenshotted into Slack, paste it into Comma instead, and send the link.

Free tier is enough for three commenters per report. No card.

Create your first report →

Related

• How to share an HTML report
• Comment on an HTML report
• Google Docs for HTML
• Where should my AI agent post its work?