A small, T-shaped team
The reverse engineering process is designed for a small, T-shaped team where members flex across roles as needed. “T-shaped” means each person has a primary specialism (the vertical bar of the T) but is willing and able to contribute outside that specialism when the work demands it.
This is a deliberate choice. The AI tooling handles much of the heavy analytical work — reading code, interpreting screens, structuring outputs — so the team can stay small and focus on steering the process, validating results, and engaging stakeholders.
Core roles
The team comprises three core roles:
1. Delivery Lead
The Delivery Lead coordinates the project, manages stakeholder engagement, ensures the process runs smoothly, and tracks progress. They are responsible for:
- Planning and scheduling the work across all phases
- Coordinating with programme product managers and application stakeholders
- Removing blockers and managing risks
- Ensuring the PRD reaches sign-off on time
2. Developer
The Developer runs the AI tooling, reviews technical outputs, and troubleshoots issues. Pairing is optional but preferred for knowledge sharing — having two developers work together reduces the risk of a single point of failure and improves the quality of technical review.
The Developer is responsible for:
- Setting up and configuring the AI tools
- Running the tooling against source code and screenshots
- Reviewing and correcting the AI-generated outputs for technical accuracy
- Escalating anything that looks wrong or incomplete
3. BA/UR (Business Analyst / User Researcher)
The BA/UR conducts stakeholder interviews, reviews curated outputs for domain accuracy, and validates the PRD against known business context. They are responsible for:
- Planning and conducting interviews with application users and product owners
- Capturing workflows, workarounds, pain points, and tacit knowledge
- Reviewing the AI-curated outputs to check they make sense from a business perspective
- Ensuring the final PRD accurately reflects how the application is used in practice
Contrast with traditional approaches
In a non-AI project, you would need a larger team with dedicated analysts spending weeks manually reviewing code, documenting screens, and writing requirements. A traditional reverse engineering effort might require:
- Multiple business analysts to manually map out each module
- Technical architects to read and interpret the codebase line by line
- Dedicated writers to produce the requirements document
The AI tooling handles much of this analysis automatically, so the team stays small. The human roles shift from doing the analysis to reviewing, validating, and enriching the AI-generated outputs.
Role activities by process phase
Each role contributes differently across the phases of the process:
| Phase | Delivery Lead | Developer | BA/UR |
|---|---|---|---|
| Gather Inputs | Coordinates access to source code and stakeholders | Sets up tooling, ingests source code and screenshots | Conducts and records stakeholder interviews |
| Content Curation | Monitors progress | Runs the AI tooling, reviews outputs | — |
| Review Curated Outputs | Facilitates review sessions | Reviews technical accuracy | Reviews domain accuracy and business context |
| Analysis & PRD Generation | Monitors progress | Runs the AI tooling, reviews outputs | — |
| PRD Review & Sign-off | Coordinates the review with stakeholders | Supports technical queries | Validates business requirements, supports sign-off |
All three roles collaborate closely during the PRD Review & Sign-off phase to ensure the final document is accurate, complete, and ready for stakeholder approval.
