What we build with it
- Services — APIs and background processing
- Automation — the work between systems
- Data work — pipelines, analysis, quality checking
- AI — retrieval, extraction, classification, evaluation
Python is the right tool for all four, which is why most of our data and AI work is written in it even where the surrounding application is not.
We measure before we build
- How often does this happen, and how long does it take?
- What does it cost now, honestly?
- What would the saving actually be redeployed to?
- What are the awkward cases?
Two weeks of measuring the current process before building costs almost nothing and it is the only baseline you will get.
Four things we insist on
- Your repository, your credentials
- Tests around the transformations and the failure paths
- Monitoring, including alerting on absence
- A named owner in your business
What we recommend against
- Automating something below the volume that repays it
- Machine learning where rules or statistics would do
- Building what a product already does well
- Automating the exception cases rather than the routine ones
- Anything nobody will own after we leave
How we communicate
A weekly note on the same day: what moved, what did not, what we need and whether the date has changed. If something is going wrong you hear it that week.
We would rather be kept because the work is good than because leaving would be difficult.