Blacknode – Visual workflow builder Claude can drive via MCP
Agents build their own workflows through typed MCP tools instead of guessing fragile JSON graphs.
Pipevals is the visual pipeline builder for evaluation-driven AI development. Build evaluation graphs. Run them against datasets. Track quality over time.
Early learning project in a crowded eval space dominated by LangSmith and Arize.
ML engineers, AI product teams building evaluation workflows
LangSmith · Arize Phoenix · Braintrust
It currently lets you: - build evaluation pipelines as graphs - run them against datasets - track how output quality changes over time
Agents build their own workflows through typed MCP tools instead of guessing fragile JSON graphs.
Hugging Face but organized by use case instead of architecture, with model comparisons.
LLM-as-judge metrics beat guessing chunk sizes, but Ragas and LangSmith already exist.
Pretty graph DB GUI, but HelixDB adoption is niche and unproven market demand.
Dynamic Pydantic models beat manual schemas for messy API responses.
83k LOC is impressive, but chaining LLMs to evaluate LLM output isn't novel architecture.