Project build.
Spec to handoff. Your team running it after we leave.
Most AI projects start as a script and end as an internal demo. We design and stand up the version that runs in production. Twelve to twenty weeks. We come in with a thesis, do the discovery, design the system, build it, ship it, and hand it back to your team operating it.
This is not staff augmentation. We will only take engagements we can lead, and we will only build production systems. The first phase exists in part so you can decide to stop. About one in five of our discovery engagements end there. That is a good outcome.
When this fits
- You have a defined business problem and a budget, and you need delivery.
- You have prototypes that did not make it to production. Most do not.
- You need an outcome, not a permanent staff augmentation contract.
- Your team is capable, but does not have the senior AI engineering muscle to lead the build.
When it does not
- The problem is not defined yet. Start with discovery as scope, or an architecture review.
- You want a software development shop. We are not one.
- You need someone to manage your existing team over time. See engineering leadership.
- Phase 01
Discovery
Weeks 1 to 4. Workflow audit. Data readiness assessment. Success metrics defined without mentioning technology. Cost model at production volume. Compliance review where it applies. By the end of this phase you can decide to stop.
- Phase 02
Architecture
Weeks 5 to 6. Selection of stack components: models, retrieval pattern, tool design, routing layer, integration points, observability. A written architecture document. Cost and reliability budgets. Risks called out in advance.
- Phase 03
Build
Weeks 7 to 14. Senior operators do the work, with your team in the room. Daily commits. Weekly demos. Production-grade code. We do not ship prototypes.
- Phase 04
Handoff
Weeks 15 to 18. Documentation. Runbooks. Operational metrics in place. Two weeks of joint operation. Then your team owns it. Thirty days of post-handoff support.
- A production system, running, with your team operating it.
- A written architecture document that explains every decision.
- Runbooks for the failure modes we found and the ones we anticipate.
- A trained team. Your engineers paired with our operators throughout.
- Thirty days of post-handoff support.
- Owns Architecture
Henrik Soerensen
Principal AI Solutions Architect
Vice President of Engineering at MGM+ and Amazon Prime Video. Led a 34-person organization at 99.999% uptime.
Read profile ↗ - Owns Correctness
Jay McCarthy, PhD
Chief Scientist
PhD in Computer Science, Brown. Co-creator of the Racket programming language. 32 peer-reviewed publications.
Read profile ↗ - Owns Integration
David Runion
Principal Engineer, AI Integrations
Eighteen years of healthcare integration work. HL7, FHIR, certified EHR systems. Ships open-source AI agent tooling: `pad` and `rome-x402-mcp`.
Read profile ↗ - Owns Experience
Angela María Cañón Piñeros
Director of AI Service Design
MSc in Communication Design, Politecnico di Milano. Thirteen years designing service experiences across Latin America, Europe, and South Asia. Trilingual.
Read profile ↗ - Owns Adoption
Marlene J. Colón Torres, PhD
Director of AI Adoption
PhD in industrial-organizational psychology. Seventeen years of organizational development across private, public, and non-profit sectors.
Read profile ↗
From $100,000. Scoped after the discovery phase. Most engagements land between $150,000 and $400,000 depending on the breadth of the system, the data readiness, and the integration surface.
Tell us what you are looking at. We will tell you whether this is the right engagement, a different one, or none of them.
Direct line.
info@swenor.us