Test a new capability
Find out whether a model or integration can meet a product requirement, such as quality, latency or local execution. Build a prototype or run an experiment against agreed criteria.
E.01Engagement$17,250 USD · Two weeks · Fixed scope
Engineering
Sprint
Move past the technical obstacle blocking your next build.
For product teams building AI features, developer tools or demanding software. We scope one technical obstacle, implement or test the approach, and hand over the code, results and next steps.
Fixed fee · Scope agreed before booking
$17,250 USD
$8,625 to book. $8,625 on delivery of the agreed code, checks and handoff. The scope and acceptance criteria are agreed in writing before booking.
Ten business days, starting once access and inputs are ready. One workflow, one primary codebase and one environment. Larger builds are quoted separately.
After the sprint
Continue with a separately scoped build, or hand the work to your own team.
E.02Ways to use the sprintOne obstacle per engagement
What’s blocking your next build?
Each $17,250 Engineering Sprint focuses on one obstacle. Choose one implementation or one bounded experiment.
Build a better workflow
Replace a fragile script or manual process with a focused CLI, integration or internal tool your engineers can use and maintain.
AI Reliability Sprint
For a working AI feature that fails on real inputs. Establish a baseline, map the failures, test targeted changes, and hand over reusable evaluations.
Same $17,250 fee and scope limits. Bring representative inputs, access to the system and a technical owner. Agree the quality, latency or cost question before starting.
E.03ScopeAgreed before booking
A small scope. A specific result.
Included in the sprint
- One workflow, one primary codebase and one execution environment.
- At most one new external integration, using ready inputs.
- One specified implementation or an experiment with up to two agreed approaches.
- Three to five acceptance criteria, code, repeatable checks and a handoff.
Work with a separate scope
- A complete product, multiple platforms or a broad migration.
- Large data preparation or open-ended model training.
- Production rollout, ongoing operations or on-call support.
- Additional workflows, integrations or requirements after booking.
For an implementation, the agreed checks define completion. For an experiment, completion means running the agreed methods and delivering the results, even if the desired performance target is not met. We choose the mode before booking.
Bring usable code or inputs, access to the agreed environment and one technical owner available for reviews. Dates are confirmed once these prerequisites are ready.
E.04TimelineTen business days
Three live reviews. Focused engineering between them.
We build and test the agreed approach, with asynchronous updates between reviews. The two-week delivery window starts once the prerequisites are ready.
Before kickoff
Ready to start
Scope agreed; inputs and access ready. Dates confirmed.
Readiness check
Day 1
Kickoff
Confirm the plan and acceptance checks together.
60 min · Live
Day 5
Midpoint review
Review a working change or early results and resolve decisions.
30 min · Live
Day 10
Handoff
Run through the code, checks, limitations and next steps.
60 min · Live
E.05What you getFour deliverables
Code, tested results and a clear next step.
Deliverable01
Scope and success criteria
A written scope with one technical question, explicit boundaries and three to five observable acceptance criteria, agreed before booking.
Deliverable02
Implementation or experiment
The implementation checked against its agreed criteria, or the experiment run against its agreed methods and cases. Code and recorded results show the outcome.
Deliverable03
Tests and handoff
Test code, setup instructions and implementation notes, with a walkthrough. Includes five business days of handoff questions, up to four hours of setup help.
Deliverable04
Plan for the next build
A recommendation for the next release, further engineering or a change of approach, with remaining risks and estimated effort.
E.06After the sprintChoose the next step from the evidence
If the evidence supports a build
Continue the build
Scope the integration, hardening or rollout work from the results. Continue with Odd Machine or use the handoff with your own engineers.
If the approach needs to change
Change course with a reason.
Use the tests and findings to revise the scope, try a simpler approach or stop an unpromising direction. Your team leaves with the evidence behind that decision.
E.07Questions(10)
Practical questions.
The best fit is a technical founder or product team with one important obstacle in an AI feature, developer tool or software product. You have usable inputs, a technical owner and a concrete decision or deliverable. We check that the work fits the sprint before you book.
