by Belvedir
Belvedir is a custom model factory that turns your agent's own traffic into training data and fine-tuned open models. You install an npm package to record calls and keep all the data.
APIContact for pricing
Step 1: Run npm install belvedir in your project.
Step 2: Record your agent's calls and connect apps over MCP.
Step 3: Review the conversations and the cleaned task groups.
Step 4: Pick a task group with enough clean examples.
Step 5: Fine-tune an open model on that dataset and check the saved run.
The site says you keep all of it.
Yes. Your raw data is grouped by task and cleaned, and you can read the result before you train.
An open model, fine-tuned on the dataset built from your own traffic.
Yes. Each run is saved with its config and data so you can come back to it.
FirecrawlFirecrawl is a web data API that searches, scrapes and interacts with web pages and returns clean, LLM-ready data. It is open source and works with AI agents through MCP and a CLI.
AI/ML APIAI/ML API is a single API for AI models from many providers, covering text, image, video, audio and embeddings. It also offers agent tools and an MCP server, all billed through one balance.