Insights Writer

An analysis is only useful when someone acts on it. Your manager does not want forty tables. They want the three things that matter and what to do about them. The Insights Writer builds that report. It takes the data profile, the findings and the charts the rest of the Upfyn Analyst Team produced, and writes a short document a reader can finish in about five minutes.

Think of it as the editor at a newspaper. The reporters have gathered the facts. The editor puts them in order, cuts the padding, leads with the headline and makes sure every picture matches the story. Every number in the report traces back to the analysis that produced it, so the reader can trust what they read and you can defend it in a meeting.

It also delivers. It can export a project as one JSON file that another app can read, send a short update to your messaging app, or post a notice to a web address when an analysis finishes. Like a courier who waits for a signature, it sends nothing until you say yes. You see the exact message and the destination first, and each yes covers one send.

• Write the analysis report: a short summary, an overview of the data, the key findings with their charts, and what they mean for the business.
• Lead with the numbers. Each finding is one headline with its figure, the direction of change and the implication, so a busy reader gets the point in the first line.
• Keep the report honest. Every number traces back to the analysis, and the report says plainly what the data could not show.
• Export a project's results as one JSON file, bundling the profile, findings, charts and report, so another app or a developer can read it without opening Upfyn.
• Send the top insights or a short update to your connected apps, such as Slack or Gmail, and put results into Google Docs, Google Sheets or Google Slides once you connect them.
• Set up a web notice (a webhook, which is a small message posted to an address you choose) that fires when an analysis finishes, so your own system knows the result is ready.
• Remember who your reports are for, the format they like and where results usually go, so the next report starts from the right shape.
• Work in the files, artifact and browser panels, so you can open a report, check it and share it from the same place.

Nothing leaves the project without your yes. The Insights Writer never puts a password, key or token into a report, an export or a message, and it never guesses a number it did not find.

The Insights Writer has five skills. You can ask for any of them in plain words; you do not need to know the skill's name.

report writes a publication-ready report from a data project. It covers the summary, the data overview, the findings with their charts, the recommendations and a short note on method and limits. Use it for "Write up the findings for my manager."

api bundles a project's analysis into one structured JSON file: the profile, insights, charts, report, dashboard and cleaned files. Think of it as packing the whole kitchen into labelled boxes so another team can unpack it in the right order. Use it for "Export the results so our app can read them."

live-update sends a project's top insights, report link or dashboard link to your messaging apps after you have seen the message and said yes. Use it for "Send the top three insights to our Slack channel."

notify sets up a webhook that receives a short JSON notice when an analysis finishes. It is the doorbell for your own systems. Use it for "Tell our system when the analysis is done."

pipeline-checks is a set of optional checks that run before and after an analysis: a data check first, a run-history log, and a webhook notice at the end. Each one runs only when you approve it. Use it when a report feeds a regular process and you want a safety check in place before the numbers go out.

Here is a common job, from a finished analysis to a report your manager can read on the train.

1. You ask: "Write up the findings for my manager."
2. The Insights Writer reads the data profile, the findings and the chart list. It checks that every chart it plans to mention exists in the charts folder, the way a picture editor checks every photo before the page goes to print.
3. It writes a short summary of three findings. Each one has the number, the direction of change and what it means in one plain sentence.
4. It adds a data overview: the files used, the number of records, the date range and the quality score from the Data Engineer's profile.
5. For each key finding, it places the chart and writes two or three sentences of plain explanation underneath, so the picture and the words say the same thing.
6. It writes the recommendations, ordered by business impact, with the reason for each one and the metric to watch.
7. It adds a short methodology section that lists the tools, the assumptions and the limits of the data, so nobody is surprised later.
8. It asks before it sends anything. It shows you the exact message, such as a summary for your Slack channel, and where it would go. Nothing is sent until you say yes.
9. You get a report you can read in five minutes, with its charts, ready to open in the artifact panel and share. If you also want a JSON export or a Google Doc, ask for it, and you get that too.