What we test for
Typical work
- Agents wired into the systems you already run, not a chat window next to them
- Tool layers with permissions scoped per action
- Long-running jobs that survive a restart and can be resumed
- Traces and replay, so a bad run can be explained after the fact
The coding test
Build an agent that completes a multi-step task against two tools we provide, one of which fails intermittently. It has to finish, give up cleanly, or ask, inside a step budget. Then show us the trace of a run that went wrong and say where you would have cut it off. Scored on failure handling, not on the happy path.
Five interview questions you can reuse
We use questions like these in our soft-skills and technical rounds. Steal them for your own interviews.
When is an agent the wrong answer?
A strong answer covers: When the steps are already known: write the code. An agent is for a path that is genuinely unknown, and it still needs a step budget, a timeout and an audit trail, because a loop that fails quietly is a bill nobody is reading.
A tool call fails halfway through a ten-step task. What happens?
A strong answer covers: It depends whether the step was idempotent. Retry what is safe to retry, checkpoint state so a restart does not repeat side effects, and surface a clean failure instead of letting the model improvise around a broken tool.
How do you stop an agent doing something expensive or destructive?
A strong answer covers: Permissions on the tool, not instructions in the prompt. Least privilege per action, a confirmation step for anything irreversible, a hard ceiling on steps and cost, and a kill switch that does not depend on the agent cooperating.
How do you debug a run that went wrong three steps in?
A strong answer covers: A trace holding every prompt, tool call and result, replayable against the same inputs. Without recorded traces you are guessing, because the run is not reproducible on its own.
How do you test something non-deterministic?
A strong answer covers: Fix what you can: seeded inputs, recorded tool responses, and assertions on the outcome and the side effects rather than on the wording. Then a scored set of end-to-end runs where you watch the pass rate over time instead of demanding green.
One monthly price per developer. Everything in it.
No deposit, no minimum term, no placement fee. You pay one invoice a month for each developer on your team.
Per developer, all-in. Your quote arrives by email with your exact number after you send the brief.
- Developer pay, local payroll and taxes
- Vetting: coding test, soft skills, English
- Free replacement within 90 days
- A direct line to our CEO
What changes the price
- Seniority and role. Senior engineer, tech lead or architect.
- Stack. Some stacks and niche tools cost more to staff.
- Team size. Larger teams get a better monthly rate per seat.
- White-label. Agencies reselling under their brand get partner pricing.
Terms in plain words
- First 2 weeks risk-free: if it isn't working, you don't pay for them
- Month to month, 30 days' notice to end
- No deposit and no minimum number of seats
- Want to hire the developer full-time? Buyout of 3 months' fee in year one, 2 months' after
Questions tech leads ask us
How senior are your AI agent developers?
Six or more years as software engineers, and at least one agent that ran against production systems with real permissions. A demo agent takes an afternoon; what we look for is what they changed after one did something they did not expect.
Can I see the coding test results?
Yes. Every profile card carries the score and the reviewer notes.
What does it cost?
$6,900 per developer per month, all-in, and flat: the same rate whatever the stack. Your exact quote arrives by email after you send the brief.
Skip the sales rep. Talk to the CEO.
Tell us what you're building, your stack and when you need people. Three matched profiles within 48 hours, and a straight answer on whether we're the right fit.