Plate 00
Este agente
This portfolio's agent: it understands your project and returns an initial scope with a timeline.
Problem
People who arrive with a project idea want to know quickly whether it makes sense and how long it takes, before booking a call.
Solution
A conversational agent asks a few questions, builds a structured scope and shows a timeline range computed by a deterministic table. If it makes sense, the contact reaches Lucas by email with the summary ready.
Decisions
- 01Provider-agnostic: the model is an environment variable; switching LLMs touches nothing else.
- 02The LLM neither computes timelines nor quotes prices: the timeline comes from a table, and a filter strips any money value from the stream.
- 03No tool has side effects: the lead is only sent when the person fills in the form and consents.
Stack
- Next.js
- Vercel AI SDK
- TypeScript
- Zod
- Upstash Redis
- Resend