Vellum - Why We Invested

GenAI is unlocking a new generation of enterprise applications that are smarter, more intuitive, and capable of solving problems traditional software can't touch. But as forward-thinking teams work to develop these capabilities in house, they're quickly discovering that building with GenAI is unlike anything they've done before. 

Unlike traditional software development (which relies on deterministic systems, structured codebases, and well-established practices to ensure everything works as expected), GenAI introduces a new paradigm. The logic lives in prompts rather than code, the same input can yield different outputs, and behavior can subtly shift over time. It demands a new level of observability, evaluation, and iteration. 

And that shift in how GenAI products are built has left many teams stuck in experimentation mode.  They're eager to move fast, but unable to ship reliable, production-ready systems because of a common set of challenges:

• The tools are fragmented: Teams are stitching together point solutions for prompt management, testing, deployment, and monitoring — without a single source of truth.
• The systems are brittle: A small change to the prompt can break functionality, and without proper version control or observability, it's hard to pinpoint or fix the issue.
• The outputs are unpredictable: LLMs are non-deterministic, so the same input can yield different outputs, making it difficult to test, trust, and troubleshoot AI behavior.
• The process is guesswork: Without structured evaluation, most teams rely on subjective "vibe checks" to assess progress.

In short, there’s no modern software engineering equivalent for GenAI.

No CI/CD, no unified tooling, no guardrails. Just duct tape and hope.

Vellum changes that. 

It's the end-to-end platform we wish every product and engineering team had from day one — a unified system for designing, testing, deploying, and monitoring GenAI applications. With Vellum, engineers and domain experts can easily collaborate in one place, evaluate application performance, and ship AI features into production. 

That's why leading companies like Drata, Swisscom, Redfin, and Headspace are already using Vellum to power mission-critical AI initiatives ranging from secure, compliance-ready automation to customer-facing features that ship fast. 

And it’s why we’re so proud to lead their $20M Series A, alongside our partners at Socii Capital, Y Combinator, Rebel Fund, and Eastlink Capital.

We’ve seen a lot of infrastructure companies — and a lot of founders — but few move with the velocity, clarity, and precision of this team. Akash Sharma, Sidd Seethepalli, and Noa Flaherty have built a platform that is poised to become essential infrastructure. And they’ve done it while executing a smart, inbound-driven GTM motion that’s quickly earning the trust of serious enterprise customers.

Nearly every company will build or buy GenAI solutions. For those who build, Vellum is the platform that makes it possible.

We’re excited to partner with the team as they scale into new markets and become the backbone of enterprise-grade AI development.

Read the open letter from Akash to learn more.

‍