// Delivery process

Four phases.
Predictable outcomes.

An engagement model designed for engineering leaders who answer to a board: defensible scope, realistic timelines, named deliverables, and a clean handover. No open-ended retainers. No surprises in month six.

01ASSESS
Assessment
Understand the gap before committing the budget.
1–2 weeks

A structured audit of your current platform, AI ambitions, team capability, and business constraints — sized to inform an investment decision. Output is an honest maturity scorecard, a written gap analysis, and a defensible go/no-go recommendation. Available as a standalone engagement with no obligation to proceed.

We examine
  • Infrastructure, CI/CD, and tooling audit
  • Developer experience survey (DORA baseline)
  • AI workload requirements & readiness
  • Security, compliance & cost posture
  • Team skills vs. platform engineering requirements
You receive
  • Platform Maturity Scorecard
  • AI Readiness Assessment
  • Current-state architecture diagram
  • Prioritised gap analysis & risk register
  • Honest go/no-go recommendation
📄 Assessment Report 📊 Maturity Scorecard 🤖 AI Readiness Report ⚠️ Risk Register
02PLAN
Planning
A defensible plan you can take to the board.
2–3 weeks

We design your target-state platform with the AI layer integrated from the start — not as a future phase. Tooling decisions, governance, and a cost model that includes GPU and AI spend. Nothing gets built until you've signed off on scope, timeline, and budget.

We design
  • Target architecture (platform + AI layers)
  • CNCF tool selection with rationale (ADRs)
  • AI infrastructure design (model serving, vector DB, gateway)
  • Golden paths for standard & AI workloads
  • Phased delivery roadmap with milestones
You receive
  • Full architecture documentation
  • Technology decision log (ADRs)
  • Cost model (cloud + GPU + LLM token spend)
  • Delivery roadmap with realistic timelines
  • Estimated DORA improvement projections
🏗️ Architecture Docs 📋 ADR Log 🗓️ Roadmap 💰 Cost Model (incl. AI)
03BUILD
Build
Production-grade delivery, on a timeline that holds.
6–16 weeks

Hands-on delivery in your environment, alongside your team. Platform foundation first, AI capability integrated as a parallel track. Weekly demos give you visible progress and de-risk the engagement at every step. Everything lives in your repositories from the first commit — owned by your organisation, not by us.

Platform track
  • Kubernetes + GitOps foundation (ArgoCD/Flux)
  • Backstage developer portal + service catalogue
  • Self-service infra via Crossplane
  • Observability: Prometheus, Grafana, OTel
  • Vault secrets + OPA/Kyverno policies
AI track
  • GPU node pools + NVIDIA device plugin
  • Model gateway (LiteLLM) + serving (KServe)
  • Vector database deployment & integration
  • RAG pipeline infrastructure
  • AI observability (Langfuse / Phoenix)
  • AI golden paths inside Backstage

⚡ Honest note: Timelines are tied to scope, not optimism. A greenfield IDP for a 50-person engineering org typically runs 8–10 weeks. Adding the AI capability adds 4–6 weeks. A complex legacy migration runs 12–16 weeks. We tell you upfront — we never pad estimates to look safe, and we never compress them to win the work.

⚙️ Live IDP + AI Layer 📁 All IaC in Git 📖 Runbooks 🔐 Security Docs 📈 Weekly Reports
04ENABLE
Train & Enable
Internal capability — not a permanent dependency.
2–4 weeks

Structured enablement across the platform and AI capabilities — tailored to your team's actual roles. Operations engineers get day-2 platform depth. Application engineers get the AI patterns. Everyone leaves with the runbooks and confidence to act independently. Includes a 30-day post-handover support window — and then we're done.

Platform training
  • Day-2 Kubernetes operations
  • GitOps workflows & ArgoCD management
  • Backstage extension & catalogue maintenance
  • Incident response & troubleshooting drills
AI infra training
  • LLMOps fundamentals & model lifecycle
  • Model gateway operations & routing
  • AI observability & cost monitoring
  • Onboarding new AI services to the platform
🎓 Workshop Materials 📹 Recorded Sessions 📘 Extension Guide 🆘 Escalation Runbook 30-day support window
// Typical end-to-end — medium scope AI + IDP project

A full engagement from first call to trained team typically runs 16–24 weeks. IDP-only (no AI layer): 12–18 weeks. Standalone assessment: 1–2 weeks.

W1–2
W3–5
W6–20
W21–24
Assessment
Planning
Build (Platform + AI tracks)
Enable
★ Honest note: These are realistic estimates anchored in 18 years of delivery experience. We never commit to a timeline we can't keep — and if scope or risk changes, you hear it from us before it shows up on an invoice.
// Start with phase 01

Start with the Assessment.
No commitment beyond that.

A 1–2 week assessment gives you a clear picture of platform maturity and AI readiness — enough to make an informed investment decision before scoping anything bigger.

Book an executive briefing →