Engagement · how we work

A four-phase build,
fixed deliverables.

You already know what we build. This page is about how the engagement runs — phases, what you walk away with, and the three ways to start.

Phases · 01 → 04

The build, end to end

01

Phase 01 · 1 week

Discovery sprint

Working sessions to map the real workflow — not the org chart version. We surface decisions, exceptions, approvals, and the data that actually moves.

Exit deliverables

  • Workflow map (decisions + handoffs)
  • Risk & approval surface
  • Build proposal with scope and timeline
02

Phase 02 · 2–6 weeks

Design & build

Agents, integrations, dashboards, and permission policy implemented against your real systems. Weekly demos against the workflow map.

Exit deliverables

  • Working system on staging
  • Versioned permission policy
  • Operator runbook
03

Phase 03 · 1 week

Deploy & cut over

Move from staging to production execution nodes, validate on real traffic, and hand the operator console keys. Reversible by design.

Exit deliverables

  • Production rollout plan
  • Rollback procedure
  • Live monitoring + alerts
04

Phase 04 · ongoing

Operate & evolve

Heartbeat monitoring, weekly review of blockers and approvals, and iteration cycles as the workflow changes. You own the system; we keep it sharp.

Exit deliverables

  • Weekly ops report
  • Iteration backlog
  • On-call response SLA

Deliverables · what you own

What you walk away with

Every engagement produces the same artifacts. You own all of them — source, policy, runbook, secrets architecture.

Workflow map

The system-of-record diagram of how the work actually flows, including silent steps and exceptions.

System source

Agents, orchestration, and integration code in your repo. No black boxes, no lock-in.

Permission policy

Versioned, reviewable policy declaring what each agent can do without asking — and what always requires approval.

Operator runbook

How to start, pause, inspect, and recover the system. Written for the human at 2am, not the engineer who built it.

Secrets architecture

Per-integration credentials isolated, rotated, and scoped. Nothing logged, nothing shared across workflows.

Telemetry dashboard

Live console for tasks, agents, approvals, and blockers — the same one we use to operate it with you.

Tiers · three ways to start

Pick the entry point

1 week

Discovery sprint

Fixed fee

A standalone diagnostic. Walk away with a workflow map and a build proposal — no commitment to continue.

  • 3 working sessions
  • Workflow map + risk surface
  • Sized build proposal
  • Recorded sessions
Book discovery
Most chosen

4–8 weeks

Full build

Project-based

Discovery → design → deploy → cut over. You own a production-grade automation system at the end.

  • All discovery deliverables
  • Production system in your stack
  • Operator runbook + training
  • 30 days post-launch support
Scope a build

Quarterly minimum

Embedded operator

Monthly retainer

We operate and evolve the system alongside your team. Weekly review, iteration backlog, on-call response.

  • Weekly ops review
  • Iteration cycles
  • On-call response SLA
  • New workflow additions
Talk retainer

FAQ

Things operators ask first

Do you build in our stack or yours?

Yours. Code lives in your repo, secrets in your vault, execution on your nodes or cloud. We're allergic to lock-in.

What does 'reversible by design' mean?

Every write goes through a step that can be rolled back, every deploy has a documented rollback path, and every agent action lives in an audit log you can replay.

Can we start with one workflow and add more later?

That's the default path. The first build pays for itself by proving the loop; subsequent workflows ride the same infrastructure.

Who owns the IP?

You do — the system, the workflow map, the runbook, the code. We retain the right to reuse generic patterns, not your proprietary logic.

Next step

A one-week discovery answers most questions.

Tell us the workflow. We'll map it, size the build, and you decide whether to continue.

Book a discovery sprint