Upload a CSV file. Columns: Type, Status, Level, Tool, Internal Description, Gain Type, Estimated Gain. Entries will be added for the current director and week.
| Level 0: Waterfall / Agile | Level 1: AI Assisted | Level 2: AI Directed | Level 3: AI Delegated | |
|---|---|---|---|---|
| Description | Humans execute delivery of the SDLC with minimal or ad-hoc AI use for tasks within the SDLC | Humans execute delivery of the SDLC and apply AI to all tasks within the SDLC | Humans set intent and specifications, directing AI to execute SDLC phases and prompt for human validation | AI runs asynchronously as a system of AI agents across to execute tickets end to end, humans review exceptions and approve |
| How AI is leveraged | Sporadic task support | Consistent AI assistance used by individuals | As a team member executing work under human direction | A delivery system of agents operating in the background |
| Examples | Various tasks within the SDLC such as: writing requirements, writing code, debugging, etc. | AI suggests lines of code, explains unfamiliar modules, helps debug, or drafts tests | AI is prompted through idea clarification and design, creates specs, architects, generates code, and test | A bug is assigned to AI; it triages, replicates, builds, tests, and returns a proposed solution |
| Work unit | Tasks and features | Tasks | Features | Full build (tasks, bugs, features, maintenance) |
| SDLC process | Traditional SDLC remains intact | Traditional SDLC remains intact (faster execution) | SDLC becomes a faster guided pipeline with AI | SDLC steps collapse into an asynchronous agent loop |
| Timing | Weeks | Weeks | Days / hours | Hours (asynchronous) |
| Value measured | % of committed work delivered on time | % of tickets influenced by AI | % of lines of code built by AI | % of tickets deflected away from developers |
| Impact | n/a | Max 1x improvement | 10x improvement (with effort) | 20x+ potential ballpark improvement |