briefquote replay
Quote from a client brief without the brief leaving your laptop
The scope, hours and model outputs are fixed: they were computed once on the author's laptop. Only the rate arithmetic (contingency and blend, the same formula as quote.py) is recomputed live in your browser when you change the rate.
Scope
Complexity
Tech stack
Deliverables
Risk flags
Open questions
Estimate
On-time probability
Price
Floor
Ceiling
Honest caveats
- Three sample briefs prove nothing about accuracy. This shows what the pipeline outputs, not that the numbers are right.
- A human must review every scope and quote before it goes to a client.
- The 30 jobs behind the estimates are synthetic example data, so the numbers show how the pipeline works, not what a real freelancer's history would say.
- With only 30 rows, estimates are a rough guide and the on-time figure is clamped to 5% to 95%.
- TabPFN model weights are licensed for non-commercial use only. Selling quotes built with them needs a commercial TabPFN license, or use the baseline estimator.
- Nothing on this page calls a model. Results are pre-computed and stored in the page.
Run it yourself
- Install Ollama from ollama.com, then run
ollama pull gemma3:4b - Create a virtual environment, then run
pip install -e ".[dev,estimator]" - Run
briefquote serve --open
Built for the DEV Hacktoberfest Open-Source AI Challenge, Week 2.