A proprietary agentic pipeline, a voice-agent audit tool and open-source tooling: every piece is in real use, measured in time saved and cost.
Spec-driven, TDD pipeline orchestrating Claude agents across every role of the cycle: product management, development, scrum master, QA. Every delivery goes through review, automated tests and CI, the same validation gates as human-written code.
Jared is internal tooling: the offering is a delivery service, while the technology remains proprietary.
Four projects shipped to production
| Project | Delivered in | Squad equivalent | Gain |
|---|---|---|---|
| Internal SaaS | 35 days | 4 people × 6 months | ×5 |
| Quick-Spell-Checker | 3.5 weeks | 4.5 people × 3 months | ×8.5 |
| PtiDonjon | 27 days | 5 people × 10-14 wks | ×15 |
| EdTech mission | 3 days | 5 people × 2 months | ×20 |
Costs divided by 3 to 4 across the board.
Methodology tool in beta: voice agents conduct audit interviews, score and map business processes, then produce a Business Model Canvas per entity and a strategic debrief document.
Odile industrializes the collection phase of AI transformation audits: triangulating interviews with documents and data, delivering a prioritized build-vs-buy debrief. It's the tooling that makes a solo, 3-4 week audit possible with minimal internal load on the client side.
Open-source npm package connecting Claude to tsserver for semantic code navigation. Where the model chained tool calls to locate a symbol, a single call now suffices, equivalent to an IDE's "Go to Definition".
Benchmark · renaming a type across 118 files, 1000+ references
| Metric | Claude alone | With the MCP | Gain |
|---|---|---|---|
| Time | 160 s | 22 s | ×7 |
| Tool calls | 41 | 4 | ×10 |
| Tokens | ~6,300 | ~125 | ×50 |
| Cost | ~$2.20 | ~$0.15 | −93% |