Langbase

Langbase belongs in the catalog because it is trying to become the composable backend layer for AI products, not just a single-purpose SDK. The product surface spans agents, memory, parser, threads, tools, workflows, and SDKs, which makes it relevant for teams that want a serverless platform decision rather than another isolated point solution. That matters in stack planning because the hidden cost of many AI builds is the amount of backend glue needed to connect agent runs, memory retrieval, document processing, conversation state, and product-facing APIs.

It also deserves inclusion because the public documentation is honest about how at least part of the commercial model works. The Agent primitive pricing docs explain that requests are counted as runs, that each run is effectively an API request of up to 1,000 tokens, that Hobby includes 500 runs per month with no overage, and that Enterprise removes hard limits and bills by runs. That is not a fully expanded plan ladder across the whole platform, but it is enough to treat Langbase as a serious platform option with a real consumption model rather than vague enterprise-only messaging.