# 333 Telemetry collection is configurable down to the Python module level
**URL:** https://forum.thousandbrains.org/t/333-telemetry-collection-is-configurable-down-to-the-python-module-level/1076
**Category:** Monty Code
**Created:** [April 23, 2026, 8:50pm UTC](https://forum.thousandbrains.org/t/333-telemetry-collection-is-configurable-down-to-the-python-module-level/1076 "2026-04-23T20:50:06Z")
**Posts on this page:** 4
**Page:** 3
### Author:  [@AgentRev](https://forum.thousandbrains.org/u/AgentRev)
#### Post date: [September 14, 2026, 7:44am UTC](https://forum.thousandbrains.org/t/333-telemetry-collection-is-configurable-down-to-the-python-module-level/1076/41 "2026-09-14T07:44:49Z")
@Bryan_Leavelle The telemetry system is what comes _before_ the database; it’s a messaging framework to transmit statistics from point A to B during runtime. The database is one of many such B points. Monty uses Wandb for its database needs, which is specifically tailored to AI training. Personally, I’m not sure what tangible additional benefits would a bespoke SQL database provide, at least at this stage.
---
### Author:  [@Bryan\_Leavelle](https://forum.thousandbrains.org/u/Bryan_Leavelle)
#### Post date: [September 14, 2026, 12:57pm UTC](https://forum.thousandbrains.org/t/333-telemetry-collection-is-configurable-down-to-the-python-module-level/1076/42 "2026-09-14T12:57:51Z")
this is very funny now after looking at it. this is the database manager in me wishing everyone’s results were instantly available to everyone else using the system, that’s all. shared results is what it gets ya, with fast queries 😉
---
### Author:  [@AgentRev](https://forum.thousandbrains.org/u/AgentRev)
#### Post date: [September 14, 2026, 7:11pm UTC](https://forum.thousandbrains.org/t/333-telemetry-collection-is-configurable-down-to-the-python-module-level/1076/43 "2026-09-14T19:11:22Z")
Results being instantly available is one of the core goals of telemetry 🙂. At present, the data accumulates in place, and is then collected at specific boundaries (e.g. after each episode) by logger classes, who then forward aggregated data to the database and/or visualizations. We’re trying to make data reporting more proactive by having a telemetry framework to emit data live at the point of origin. This is primarily to facilitate creating visualizations (see [2026/07 - Upcoming and Ongoing Engineering Platform Changes](https://forum.thousandbrains.org/t/2026-07-upcoming-and-ongoing-engineering-platform-changes/1188), Tristan showed a nice demo). It also enables more fine-grained database logging too, so it’s not unlikely that the database portion might get some overhauls later down the road.
---
### Author:  [@Bryan\_Leavelle](https://forum.thousandbrains.org/u/Bryan_Leavelle)
#### Post date: [September 14, 2026, 8:08pm UTC](https://forum.thousandbrains.org/t/333-telemetry-collection-is-configurable-down-to-the-python-module-level/1076/44 "2026-09-14T20:08:15Z")
ok. that is awesome. thank you!
[Previous page](https://forum.thousandbrains.org/t/333-telemetry-collection-is-configurable-down-to-the-python-module-level/1076.md?page=2)