Peter Brooksbank
@peterbrooksbank.bsky.social
Current PureClarity autonomous agent setup:
1. Extract learnings (last 24hrs commits)
2. Generate daily report, currently 10 prioritised errors out of top 100 logged
3. Picks error, creates PRD
4. Iterates until done
5. Creates PR
6. Pick next error
7. Repeat until done
1. Extract learnings (last 24hrs commits)
2. Generate daily report, currently 10 prioritised errors out of top 100 logged
3. Picks error, creates PRD
4. Iterates until done
5. Creates PR
6. Pick next error
7. Repeat until done
February 5, 2026 at 12:50 PM
Current PureClarity autonomous agent setup:
1. Extract learnings (last 24hrs commits)
2. Generate daily report, currently 10 prioritised errors out of top 100 logged
3. Picks error, creates PRD
4. Iterates until done
5. Creates PR
6. Pick next error
7. Repeat until done
1. Extract learnings (last 24hrs commits)
2. Generate daily report, currently 10 prioritised errors out of top 100 logged
3. Picks error, creates PRD
4. Iterates until done
5. Creates PR
6. Pick next error
7. Repeat until done
16% savings on our infra cost from shutting down our data warehouse last week.
That's 16% of our infra cost on a feature that never got used. Several thousand £ a year.
Everything is obvious in hindsight, but know who you're building for.
That's 16% of our infra cost on a feature that never got used. Several thousand £ a year.
Everything is obvious in hindsight, but know who you're building for.
January 27, 2025 at 3:38 PM
16% savings on our infra cost from shutting down our data warehouse last week.
That's 16% of our infra cost on a feature that never got used. Several thousand £ a year.
Everything is obvious in hindsight, but know who you're building for.
That's 16% of our infra cost on a feature that never got used. Several thousand £ a year.
Everything is obvious in hindsight, but know who you're building for.
Last night we finally shut down our data warehouse, removing the code and deleting the database.
This was a really powerful feature which provided predictions to customers for how well a new campaign, segment or recommender would work.
The problem?
This was a really powerful feature which provided predictions to customers for how well a new campaign, segment or recommender would work.
The problem?
January 23, 2025 at 11:40 AM
Last night we finally shut down our data warehouse, removing the code and deleting the database.
This was a really powerful feature which provided predictions to customers for how well a new campaign, segment or recommender would work.
The problem?
This was a really powerful feature which provided predictions to customers for how well a new campaign, segment or recommender would work.
The problem?