We're excited to announce that Mastra has raised a $22M Series A led by Spark Capital, bringing our total capital raised to $35M.
18 months ago, we walked out of a NYC AI hackathon trying to figure out why people were writing Python notebooks instead of shipping TypeScript agents. The next day, we decided to start Mastra.
Great frameworks obsess about details to feel like magic
Great frameworks need great primitive design. API design. Documentation, examples, integrations, onboarding.
Great frameworks give you the tools you need (power), with the right insertion points (extensible), properly sequenced (progressive disclosure), working well together (interoperable).
Done right, great frameworks feel like magic. They make learning new domains easy (or easier). They bring devs into the future, so they don't get left behind in yesterday's stack.
Catch us at a bar and we'll talk your ear off about this stuff. We're framework guys. We love it.
Watching builders build
Every week we have hundreds of conversations on Slack, Discord, Zoom, and LinkedIn. We get to watch tens of thousands of first-time AI engineers build complex agents and applications with Mastra.
Teams at scale-ups like Brex, Sanity, and Factorial building agents within their apps, so users can accomplish tasks faster and easier.
A team at Indeed that built a nationally advertised career counselor agent.
A team at Marsh McLennan that built enterprise search used by 100k+ people every day.
The team at Brex building agents that helped drive their $5.1B Capital One acquisition.
Platform teams at companies like MongoDB, Workday, and Salesforce reducing toil by automating DevOps and SRE work.
Platforms like Replit letting their users build agents and automation.
More powerful models need more powerful harnesses
Some people argue that as models get better the harnesses will go away. But over the last 18 months, the models have gotten much better. And harnesses have become bigger, not smaller.
It's why we've shipped parallel tool calls, subagents, sandboxes, filesystems, skills as Mastra agent primitives (and we're working on heartbeat, channels, browser access, and web data).
It's why we shipped evals, logs, traces, datasets, an agent editor, and auth with RBAC in Mastra Studio (and we're working on annotations).
Starting today, Mastra is a platform, too
Mastra is a framework, and today, we're excited to launch the Mastra platform with the tools you need to run your agents effectively at scale:
- Mastra Studio is cloud-based and self-hosted: evals, logs, traces, datasets, annotations.
- Mastra Server lets you deploy your agents and workflows.
- Memory Gateway gives you state-of-the-art agent memory, whether you're using Mastra or another framework.
Putting agents in the hands of every developer
Early Internet users coined the term "Eternal September" to describe a flood of users that never stopped. In 1993, Internet usage was growing 2,300% per year.
Something similar is happening today with agents.
Dozens of devs are building agents today for every one building 18 months ago.
In another couple of years, there will be hundreds for every one now.
We're putting agents in the hands of every developer. And now, with AI, everybody can be a developer.
That's our mission at Mastra.
