SOLUTIONS THAT WORK
Enterprise engineering pilot readiness guide

Commission a representative system increment that exposes real complexity. Establish evidence for architecture, integration, reliability, and ownership while assessing delivery pace. Include agent behavior where the system uses agents.

1. SYSTEM AND TECHNICAL OWNERSHIP
[ ] Describe the custom application or enterprise agentic system.
[ ] Name the product owner and technical decision owner.
[ ] Identify complex domain rules, state transitions, and important exceptions.
[ ] Select a meaningful system slice with representative integration and operational constraints.
[ ] Distinguish a discovery/foundation phase from implementation where needed.

2. ARCHITECTURE AND INTEGRATION
[ ] Identify system interfaces, API contracts, data stores, and identity boundaries.
[ ] Where agents are needed, define tools, allowed actions, durable state, and orchestration.
[ ] Establish data handling, approved model providers, retention, and location requirements.
[ ] Confirm scoped access to repository and deployment targets.
[ ] Review existing systems and dependencies; do not assume integration feasibility.

3. ACCEPTANCE AND RELIABILITY
[ ] Define correct behavior with representative examples and edge cases.
[ ] Agree latency, throughput, recovery, and other workload-specific targets.
[ ] Define authorization, human approval, escalation, and safe failure behavior.
[ ] Choose tests and, where applicable, agent evaluations that exercise actual risks.
[ ] Agree how to review security findings, defects, and unresolved limitations.

4. ENGINEERING APPROACH AND COMMERCIAL CLARITY
[ ] Choose the expertise and engineering roles the work needs.
[ ] Agree approved AI coding tools and development practices with the client.
[ ] Decide where agents or orchestration improve the work; they are not prerequisites.
[ ] Scope tool, model, infrastructure, and support costs where applicable.
[ ] Agree scope, commercial terms, cost visibility, and change rules before work starts.
[ ] Review enterprise-grade quality and delivery speed together.

5. DELIVERY RESPONSIBILITIES
[ ] Choose custom agentic system service, embedded senior engineers, or owned delivery.
[ ] Agree scope, acceptance, dependencies, commercial cap, and change rules.
[ ] Name domain experts and security/cloud approvers.
[ ] Agree collaboration hours, evidence reviews, and decision turnaround.
[ ] Define IP rights, hosting, support, source ownership, and exit obligations.

6. EVIDENCE AND CONTINUATION
[ ] Review working software, tests, scans, release evidence, and agent evaluations where applicable.
[ ] Review delivery pace and agreed costs alongside engineering results.
[ ] Inspect observability, recovery guidance, and operating ownership.
[ ] For owned delivery, test whether a client engineer can understand, change, verify, and deploy.
[ ] Use 10-day project loops as a review cadence, not an entire-system completion promise.
[ ] After an agreed evaluation period, continue, redirect, or stop based on evidence.

Initial owned-delivery focus: AWS greenfield systems, Python APIs, JavaScript/TypeScript applications. Brownfield integration, additional clouds, and restricted environments require feasibility review.

This guide supports technical discovery. Contractual commitments, security controls, acceptance targets, and commercials must be agreed for each engagement.