Do you take equity?
No. Equity contracts complicate operating relationships and we want the operational freedom to walk away if a fit fails.
How we engage
How we engage
4
Phases against a Statement of Work. The same shape, every engagement. Senior people, clear scope, measurable handoff.
Every engagement is shaped against a Statement of Work. Specifics are agreed on the qualifying call and locked in the SOW.
01
Discovery
2 weeks02
Architecture
4 weeks03
Build
12–24 weeks04
Operate
Ongoing, 6 to 12-month contracts- 01
Discovery
2 weeks - 02
Architecture
4 weeks - 03
Build
12–24 weeks - 04
Operate
Ongoing, 6 to 12-month contracts
01
Discovery
2 weeks
Operations audit, stack mapping, opportunity sizing. Two weeks inside your operation with the people doing the work.
What we do
- Sit with the operating team for 4–6 hours total across the engagement
- Map every workflow, system, and decision point in scope
- Quantify where AI + HITL meaningfully changes throughput or accuracy
- Draft a thin architecture and identify integration risks
What you commit
- Read access to relevant systems
- 4–6 hours of operating-team time, not just executive sponsors
- Honest answers about what is broken, political, and previously tried
Deliverables
- Opportunity map ranked by impact, with sized lift bands
- Stack inventory + integration feasibility ratings
- Recommended Architecture phase scope, or honest "we are not the right fit"
Sample artifacts
- 6–10 page opportunity map document
- One-page stack diagram
- 90-minute readout with the operating team (recordable)
02
Architecture
4 weeks
Solution design, HITL touchpoints, integration plan, and a Statement of Work for the Build phase.
What we do
- Design the system end-to-end with explicit human-in-the-loop touchpoints
- Draft data flow, model selection, and approval surfaces
- Build a thin prototype on representative data to de-risk assumptions
- Produce a Build-phase SOW with named milestones and acceptance criteria
What you commit
- Decision-makers in design reviews (not just delegates)
- Sign-off on HITL touchpoints and risk model
- Procurement aligned for Build phase kick-off
Deliverables
- Architecture document with HITL surfaces named
- Working prototype on a representative dataset
- Statement of Work for the Build phase, fixed-scope
Sample artifacts
- Architecture document (typically 15–25 pages, technical)
- Prototype walkthrough recording
- Build-phase SOW with milestone schedule
03
Build
12–24 weeks
Implementation, integration, and deployment to production against the Architecture phase SOW.
What we do
- Ship the system against the Architecture SOW, milestone by milestone
- Integrate with systems of record on read and write paths
- Stand up monitoring, audit logs, and the operator review console
- Run UAT with the operating cohort that will use the system daily
What you commit
- A named operations counterpart, ~30% allocation through Build
- UAT cohort drawn from people who will use the system daily
- Production access on the agreed schedule
Deliverables
- Production system in your VPC or ours, your call
- Operator console with full HITL review surfaces
- Runbook + incident playbook
- Audit log queryable from day one
Sample artifacts
- Bi-weekly milestone demos with the operating team
- Production deployment runbook
- Monitoring + alerting set up against your channels
04
Operate
Ongoing, 6 to 12-month contracts
Tuning, monitoring, and expansion. The phase where AI either keeps earning its keep or it does not.
What we do
- Monitor accuracy, latency, and human override rates weekly
- Tune models, prompts, and approval thresholds as the operation evolves
- Run a working forum with the operating team on a fixed cadence
- Expand scope when foundations are earning their keep
What you commit
- A working forum (weekly or biweekly) with the operating team
- Visibility on metric targets and override rates
- Annual roadmap review
Deliverables
- Monthly operating report (metrics, drift, override audit)
- Quarterly roadmap update
- On-call coverage to your SLA
Sample artifacts
- Monthly report (5–10 pages, metric-driven)
- Quarterly roadmap document
- Operate-cadence Slack or Teams channel for the working forum
Frequently asked
Questions procurement asks first.
What happens if Build runs late?
Build is fixed-scope against the SOW. If we miss a milestone we keep working on the current milestone · the engagement does not move forward until it ships. We have not had to invoke this in practice, but it is the contract.
Can we hire you for just Operate, on a system someone else built?
Sometimes. We need a shortened Discovery first to assess the system and the operating context, and a clear handoff from whoever built it. We say no often enough that the asking is worth it.
Will you sign our MSA, our SOC 2 questionnaire, our DPA?
Yes, within reason. We work to standard procurement processes, sign a DPA when GDPR or CCPA applies, and complete vendor security questionnaires as part of the engagement. We will push back on terms that fundamentally conflict with how we deliver.
Where is the data?
Your call. We can deploy in your VPC (preferred when data residency or compliance posture matters) or in ours (faster to stand up). Either way, your data stays your data, no training on client data, no cross-engagement reuse.
Will you talk to our existing vendor / SI partner?
Yes. We integrate against the systems they manage every engagement. We do not compete with implementation partners on the platform layer, we sit on top.
Get started
Start with a 30-minute qualifying call.
We'll walk through the operation you're thinking about, talk through how the engagement would shape up, and either move toward Discovery or be honest about why we shouldn't.