Hi everyone,
Pulse 8 introduced an MCP server. It lets an AI assistant such as Claude query your Pulse data directly: performance history, thread snapshots, chore and process runs, alerts, change history and source code. To show what that makes possible, we pointed it at one TM1 instance and built three reports. They’re attached, with the data anonymised.
1. Lock and usage review
A two-week investigation that reconstructs a blocking chain from Pulse thread samples. One MDX query held the model for 4 hours 48 minutes. A housekeeping chore queued behind it, and a user’s hierarchy edits queued behind the chore. It also matches the lock days with model changes and shows who actually uses the engine, not just who stays connected.
demo-workforce-fpm-lock-usage-dashboard.html (29.6 KB)
2. Server documentation
How the model is wired, layer by layer, from ingest to the GL hub to the planning roundtrip. It’s built from the actual rule and TI source code, not from object names.
demo-workforce-fpm-server-documentation.html (69.6 KB)
3. A weather forecast for TM1
Current conditions, a 7-day outlook and an hour-by-hour forecast for tomorrow, each with its “showers” and “storms”. Click any item to see its run history, the thread evidence and a suggested fix. For example, it found that a nightly control-object copy waited on another chore on exactly the 13 days, out of 21, when that chore overran by more than 100 seconds, and on none of the others.
demo-workforce-fpm-weather-forecast.html (54.6 KB)
tm1-weather-forecast-SKILL.zip (4.7 KB)
Why this matters
These three reports are only examples. The Pulse MCP gives you the data, and how you present it is up to you: an audit report, a morning health check, model documentation, or something more creative. The weather forecast started as a colleague’s idea, and it turned out to be a very natural way to explain recurring load patterns to people.
We’d love to see what you build. Share your ideas, or your own reports, in this thread.
Cheers,
Vincent