Reinvent the workflow.
We look at how the work should run before we automate any of it.
Not AI for the sake of AI.
We build agents that do the work by your business rules. Your people approve anything that doesn't fit.
How we work
Build the tools.
We write the software ourselves, on the systems you already use.
Deterministic first.
Plain code for the predictable parts. AI for the messy ones.
Latest notes
All readingsThe bottleneck moved
Redesigning source control at Anthropic (opens in a new tab)
Anthropic
A hard stop on spend
We're going to need default hard budget caps on pretty much everything (opens in a new tab)
Simon Willison
Agents make it easy to start a job that keeps spending money while you sleep. Set a hard monthly cap that stops the work, and treat a warning email as too late.
A model with a computer
Apps, Agents, and Aggregation (opens in a new tab)
Ben Thompson
An agent is a model with a computer, so it can use the software the company already runs. Once that works, the scarce part is knowing which job you actually want done.
Tips
All tipsWatch it done before you automate it.
The side spreadsheet people keep is the real spec.
Test on last month's work.
If the results don't match what your team did, it isn't ready.
Start with one workflow.
Prove it there, then copy it.
Get in touch
Tell us one workflow that's eating your team's time.
When agents write most of the code, the slow part becomes source control, builds, and review. Plan for that pile of changes before you measure the project by how fast the model can type.