Vellum

Vellum is a platform for building and operating LLM powered applications, with an emphasis on agents, prompt engineering, and measurable evaluation. The product site describes creating agents by chatting with AI as well as using low code and code approaches, which helps teams move from prototypes to more structured workflows. Vellum also publishes dedicated product pages for prompt engineering and evaluations, supporting side by side model comparisons and test driven iteration so changes can be measured for regressions. On the official pricing page, Vellum offers a Free plan at $0 per month for one user with credits, hosted agent apps, a debugging console, and a limited knowledge base allowance, and a Pro plan at $25 per month that increases builder credits for more experimentation. In practice, teams get the most value by versioning prompts and workflows, running evaluations against representative datasets, and logging outputs so reviewers can spot failure modes such as hallucinations or inconsistent tool use. For deployment, hosted agent apps can simplify sharing with stakeholders, but you should define data boundaries and retention expectations before adding sensitive documents to a knowledge base. Vellum fits product, data, and engineering teams that want to ship AI features with clearer observability and repeatability than ad hoc prompt tweaking.