Process · how I ship
5 stagesHow I ship — stage by stage.
Five stages from discovery to handoff. Here is what happens in each, what I need from you, and what lands on your side at the end.
Stage overview · 5 stages · click a number to jump
Stage docket, entry by entry
5 stages · what happens · what I need · what you get
01
Discovery
Sixty to ninety minute call. I map your workflow end-to-end, ask for access to the two or three systems involved, and scope the first sprint. You get a written proposal with a fixed price inside twenty-four hours.
What I need from you
- Access to the relevant systems (Slack, Drive, repo, or equivalents)
- Two to three hours for the discovery call and a follow-up review
- A decision-maker in the room for scope sign-off
What you get
- Written proposal (PDF)
- Scoped sprint plan with milestones
- Fixed-price quote valid for 30 days
Week 01 · Day 1–3 · typically 3 hours of your time
02
Architecture
I design the agent roles, tool use, guardrails, and data model. You get a technical design doc and an architecture diagram before any code gets written. This is where I flag the hard integration questions — not after the sprint is underway.
What I need from you
- Sign-off on the sprint proposal
- Technical contact for integration-specific questions
- Existing schema / API docs where relevant
What you get
- Architecture diagram (agent topology + data flow)
- Technical design document
- Integration map with named endpoints
Week 01 · Day 4–5 · typically 1–2 hours of your time
03
Build
Iterative implementation with continuous testing. I ship working code to staging from day one, not at the end. You review progress in a weekly check-in and get access to the staging environment throughout the sprint.
What I need from you
- Staging environment access (or I spin one up)
- Weekly thirty-minute check-in for the duration of the sprint
- Fast decisions on scope clarifications as they come up
What you get
- Working system on staging (from day 3)
- Build log (public Notion, or private if scoped that way)
- Commit-level progress visibility
Week 02–03 · typically 1 hour of your time per week
04
Testing
End-to-end testing with production-like data, agent eval metrics, and UAT with a decision-maker on your side. Edge cases get named and handled; performance budgets get verified; anything that fails the eval threshold gets a follow-up sprint or a scope conversation.
What I need from you
- UAT signoff from one decision-maker
- Production-like test data (anonymised if needed)
- Named success metrics from the discovery phase
What you get
- Test coverage report
- Agent evaluation metrics against named thresholds
- Edge-case handling documentation
Week 03 · typically 2–3 hours of your time
05
Handoff
Production deployment, a sixty-minute handoff call, and the full source + docs transfer. You own everything — source code, architecture docs, prompt library, training video. Post-launch support runs for one to two weeks depending on the tier.
What I need from you
- Production credentials (if not already scoped)
- A decision-maker on the handoff call
- One round of final UAT on production
What you get
- Source code + repository ownership
- Documentation + architecture + prompt library
- Training video (30–45 min walkthrough)
- 1–2 weeks of post-launch support
Week 04 · typically 3 hours of your time
What you own at handoff
5 deliverables · source · docs · training · support
01
Source code
Yours. Repo transferred on handoff. No licensing, no strings, no hidden API keys.
02
Documentation
Architecture diagram, data model, agent prompts, integration map. Everything a second developer needs.
03
Training video
Thirty to forty-five minutes walking through the working system — narrated, recorded, shared.
04
Post-launch support
One to two weeks of included fixes + tuning. Bug reports get priority; new features get scoped as a follow-up.
05
Handoff call
Sixty minutes on a call with your decision-maker to transfer the mental model and close the engagement.
Process questions
Questions I get asked before starting.
How long does a typical project take?
Sprint engagements deliver inside 1–2 weeks. Full builds take 2–4 weeks. Ongoing retainers operate on a monthly cycle, with one capability sprint per month plus monitoring.
What do I need to provide during the build?
Access to the workflows being automated, two to three hours for discovery, and a weekly thirty-minute check-in. All the technical work lives on my side.
What happens if the agents do not perform as expected?
Every engagement includes post-launch support — one to two weeks depending on tier. I monitor performance, tune prompts, and iterate until the system hits the thresholds named in the discovery phase.
Do I own the code and the agents after the project?
Yes, one hundred percent. Full source code, documentation, architecture diagrams, and training. No vendor lock-in, no ongoing licensing fees, no hidden API-key dependencies.
Can the process be customised for my situation?
The sprint framework is a starting point, not a contract. Timelines, deliverables, and involvement levels adapt to your situation. I will flag the adjustments at the discovery call and quote the sprint around them.
What if I need changes after handoff?
Three options. Iterate yourself using the documentation, engage me on the ongoing retainer, or book another sprint. The system is designed to be extensible either way.
Verdict · audited 2026-04-20 · 5 criteria
Discovery week 1, sprint weeks 2–3, handoff week 4.
Sprint engagements collapse architecture, build, and testing into a single week. Full builds stretch build and testing across two weeks. Ongoing engagements re-enter the build stage on a monthly cadence. Discovery and handoff never get skipped — the proposal and the handoff call are where scope creep dies.
- Audited
- 2026-04-20
- Criteria
- 5
- Status
- on the books
- Voice
- first-person
Start with discovery.
I take on two or three projects a month. Tell me what you need and I will scope the first sprint.