Define the smallest useful result
On the fit call we identify the workflow, deadline, systems, and acceptance test, including the data-residency and traceability controls it has to satisfy. You get a direct answer, not a pitch deck.
For legal-tech teams with a deadline and an open AI role
Your legal AI hire may not land before the release.
Your legal AI hire may not land before the release.
We can ship the feature while you keep hiring.
Give us one legal workflow - document retrieval with citations, contract review, obligation extraction. We define the acceptance test, build it in your environment with traceability designed in, and hand an operable system to your team. Technical scoping or a prototype starts at EUR 1,000; production builds are fixed-price after scope.
Check if your feature fits
30 minutes. Leave with a yes, a no, or a tighter scope.
Fixed scopeBuilt in your accountsTraceable by designClean handoff
Start with one deliverable, not a large commitment
A technical scope, evaluation setup, integration, or prototype can start at EUR 1,000. If the feature needs a production sprint, you approve a fixed deliverable, price, and acceptance test before work begins.
Watch it work every week
The build runs in your environment. Weekly demos expose progress, tradeoffs, and failures before handoff day.
Own the result
Your team keeps the code, infrastructure, evaluations, audit trail, deployment notes, and recorded walkthroughs. No required retainer.
This is for you if
- You are recruiting a senior AI engineer, but a customer release cannot wait for the search to close.
- A retrieval or document-review prototype works in a demo but still needs citations, evaluations, traceability, and failure handling before production.
- The workflow touches sensitive legal data and needs data-residency and confidentiality controls.
- Your team can maintain the feature after launch but cannot absorb the initial delivery work right now.
Not sure the scope is small enough? Book the call. We will recommend a smaller first deliverable or tell you plainly that we are not the right fit.
Meet the builders. Inspect the delivery standard.
We will not manufacture customer quotes. Instead, you can inspect the people leading the work, their public technical writing, and the acceptance criteria used to define a real result.
Dmitri Bogatenkov
AI and systems architecture
Production AI, RAG, evaluation, and HIPAA and SOC 2 workflow architecture.
Sergei Bogatenkov
Product design and delivery
More than a decade in software, plus AI-native interfaces, integrations, and product delivery.
PassAnswers cite only permitted source documents and case law
PassHuman approval is required before any external filing or action
PassEvery run records latency, model cost, and a traceable citation
PassFailure paths return a specific recovery action
Targets are adapted to your workflow and signed off before delivery begins.
What can fit
The best first engagement has one user, one workflow, and one result your team can verify.
Good starting points
- Legal-search RAG with citations and traceability
- Contract or document review with human approval
- Clause or obligation extraction with an audit trail
- On-prem or data-residency retrieval workflow
Usually not a fit
- Replacing your entire case-management platform in six weeks
- Open-ended "AI transformation" without a user workflow
- Five unrelated features presented as one project
- A build without access to data, systems, or an internal owner
Inside a full production sprint
Smaller starting engagements use only the steps they need. For a six-week production build, the work advances through this sequence with a working demo every week.
Week 1
Scope, data, architecture
Scope sign-off, data access, model selection, integration map, data-residency controls, and a working development environment.
Weeks 2-3
Working build and integration
The core workflow runs end to end, then connects to your document stores, auth, and existing systems.
Weeks 4-5
Evaluation, security, observability
Edge cases, evaluation targets, latency and cost limits, access controls, audit logs, and monitoring.
Week 6
Launch and handoff
Production deployment, documentation, audit trail, recorded walkthrough, and a clean transfer to your team.
Questions we get on the scope call
How do you handle citation accuracy and traceability?
We build retrieval that cites only permitted sources, records a traceable citation for every answer, and adds human approval before any external action. Data-residency and access controls are designed in.
How can you ship a real legal AI feature in 6 weeks?
We limit the engagement to one bounded workflow, define acceptance criteria before work starts, build in your environment, and demonstrate working software every week. Broader products are split into smaller deliverables instead of forced into an unrealistic deadline.
Do you work inside our confidentiality and data boundaries?
Yes. The build runs in your accounts and environment, with GDPR, data-residency, and confidentiality requirements designed into the feature rather than bolted on afterward.
Who specifically leads my sprint?
One of the co-founders, Dmitri or Sergei Bogatenkov, is on every sprint as the technical lead. Dmitri's background is production AI and systems architecture. Review Dmitri's profile or Sergei's profile.
What happens after the sprint?
You own the code, models, infrastructure, deployment, evaluations, and documentation. We can support the handoff or discuss follow-on work separately, with no retainer required.
How much does it cost?
Technical scoping, an evaluation setup, or a prototype can start at EUR 1,000. Larger production builds receive a fixed scope, price, and acceptance test before work begins.
Take one legal-AI feature off the hiring critical path
The 30-minute call ends with a yes, a no, or a smaller recommended scope. Technical starting deliverables begin at EUR 1,000, with no required retainer.
Not in legal tech? See the main AI Feature Sprint.