Excited to share that @smthomas3, Abhi Aiyer and I are building Mastra, a Typescript AI framework for the next million AI developers:

@devagrawal09 Engineers are very bad at reasoning about pricing

@lit_lithium @ycombinator @mastra_ai @abhiaiyer called our shot

@pitdesi Sure. But was pretty confusing at the time to have two competing HR / payroll systems that sounded similar.

@pitdesi ZenPayroll => Gusto also comes to mind

@abhiaiyer 🔥🔥🔥

@resend @AnthropicAI us :)

@abhiaiyer @ycombinator @mastra_ai 🔥🔥🔥

Here's the full changelog:

Mastra now uses LibSQL as the default storage primitive. Here's how to use it:

Mastra has added a bundling system using Rollup with tree shaking enabled to eliminate unused code across packages. In order to take advantage of this, we recommend using the recommended import syntax. This has helped us reduce the core bundle size to 270kb from around 1MB.

Mastra now has short-term working memory. It uses XML-like tags to track and update contextual information during conversations. Agents can maintain persistent information through tagged data in a conversation stream

Mastra now uses AI SDK directly for model routing, removing our wrapper implementation. You can import `embed` and `embedMany` straight from `ai-sdk` And add `ai-sdk` models to Mastra agents:

@staysaasy 🤣🤣🤣

@willwill_s @emollick That is also a plausible possibility.

@justinlebar @emollick I agree with this, but do notice that folks are pretty comfortable making predictions about events. I think it’s harder to make predictions about how people’s lives will change