The example from the landing page does not exactly spark joy:<p><pre><code> testWorkflow
.step(llm)
.then(decider)
.then(agentOne)
.then(workflow)
.after(decider)
.then(agentTwo)
.then(workflow)
.commit();
</code></pre>
On a first glance, this looks like a very awkward way to represent the graph from the picture. And this is just a simple "workflow" (the structure of the graph does not depend on the results of the execution), not an agent.
I get the same feeing when I first looked at the LangChain documentation when I wanted to first start tinkering with LLM apps.<p>I built my own TypeScript AI platform <a href="https://typedai.dev" rel="nofollow">https://typedai.dev</a> with an extensive feature list where I've kept iterating on what I find the most ergonomic way to develop, using standard constructs as much as possible. I've coded enough Java streams, RxJS chains, and JavaScript callbacks and Promise chains to know what kind of code I like to read and debug.<p>I was having a peek at xstate but after I came across <a href="https://docs.dbos.dev/" rel="nofollow">https://docs.dbos.dev/</a> here recently I'm pretty sure that's that path I'll go down for durable execution to keep building everything with a simple programming model.
Kind of similar camp, I checked LangChain and others and ultimately I was like, well, it's not really doing much is it, just adding abstraction on top of what is essentially basic loops and conditional statements, and tbh it feels like in nearly every case I'll never be using them the same way such that some abstraction will help over just making some function helpers myself.<p>I don't think from first principles there's any broad framework that makes sense to be honest. I'll reach for a specific vector DB, or logging library, but beyond that you'll never convince me your "query-builder" API is going to make me build a better thing when I have the full power of TypeScript already.<p>Especially when these products start throwing in proprietary features and add-ons with fancy names on top.
TypedAI looks solid, was not aware of it! Bookmarked for further research.<p>Personally I am not fond of the decorator approach and decided to not use it in pgflow (my soon-to-be-released workflow orchestration engine on top of Postgres).<p>1. I wanted it to be simple to reason about and explicit (being more verbose as a trade-off)<p>2. There are some issues with supporting decorators (Svelte <a href="https://github.com/sveltejs/svelte/issues/11502">https://github.com/sveltejs/svelte/issues/11502</a>, and a lot of others).<p>3. I decided to only support directed acyclic graphs (no loops!) in order to promote simplicity. Will be supporting conditional recursive sub-workflows to provide a way to repeat some steps and be able to branch.<p>Cheers!
Can dbos work with CF durable objects?
Thanks! The conditional `when` clauses live on the steps, rather than being represented in the workflow, and in fact when we built this for an example, the last step being called depended on the results of the previous two steps.<p>How would you simplify this?
I think the problem is that a 'fluent' chain of calls already expresses a sequence, so the way that 'after' resets the context to start a new branch feels very awkward ... like a GOTO or something<p>It's telling that the example relies on arbitrary indentation (which a linter will get rid of) to have some hope of comprehending it<p>Possibly this was all motivated by a desire to avoid nested structures above all?<p>But for a branching graph a nested structure is more natural. It'd also probably be nicer if the methods were on the task nodes instead of on the workflow, then you could avoid the 'step'/'then' distinction and have something like:<p>e.g.<p><pre><code> testWorkflow(
llm
.then(decider)
.then(
agentOne.then(workflow),
agentTwo.then(workflow),
)
)</code></pre>
I think it is just easier to comprehend if the edges/dependencies are explicit (as an array for example).
Yeah, I also found this a bit unintuitive at first. I’m building a workflow engine myself (<a href="https://pgflow.dev/pgflow" rel="nofollow">https://pgflow.dev/pgflow</a>, not released yet), and I’ve been thinking a lot about how to model the DSL for the graph and decided to make dependencies explicit and use method chaining for expansion with other step types.<p>Here’s how it would look like in my system:<p><pre><code> new Flow<string>()
.step("llm", llmStepHandler)
.step("decider", ["llm"], deciderStepHandler)
.step("agentOne", ["decider"], agentOneStepHandler)
.step("agentTwo", ["decider"], agentTwoStepHandler)
.step("workflow", ["agentOne", "agentTwo"], workflowStepHandler);
</code></pre>
Mine is a DAG, so more constrained than the cyclic graph Mastra supports (if I understand correctly).
I knew it will be bad when I seen "by the developers of Gatsby", but this is pure comedy.<p>JQuery plugin for LLM.
Very excited about Mastra! We have a number of Agent-ic things we'll be building at ElectricSQL and Mastra looks like a breath of fresh air.<p>Also the team is top-notch — Sam was my co-founder at Gatsby and I worked closely with Shane and Abhi and I have a ton of confidence in their product & engineering abilities.
Why not use Elixir for agents as Electric is already heavily invested? It’s a much better fit than JS.
Gretchen, stop trying to make Elixir happen.
I think it is actually a solid choice given the startup ecosystem and generally easy async nature.
Abhi is one of the best engineers I know. I’m excited that he and his colleagues are tackling this problem.
This looks awesome! Quick question, are there plans to support SSE MCP servers? I see Stdio [0] are supported and I can always run a proxy but SSE would be awesome.<p>[0] <a href="https://mastra.ai/docs/reference/tools/client">https://mastra.ai/docs/reference/tools/client</a>
we have a tutorial that covers this!<p><a href="https://docs.mcp.run/tutorials/mcpx-mastra-ts" rel="nofollow">https://docs.mcp.run/tutorials/mcpx-mastra-ts</a><p>you don't even need to use SSE, as mcp.run brings the tools directly to your agent, in-process, as secure wasm modules.<p>mcp.run does have SSE support for all its servlet tools in the registry though too.
Added support in this PR <a href="https://github.com/mastra-ai/mastra/pull/1957">https://github.com/mastra-ai/mastra/pull/1957</a>! Isn't shipped just yet but will be soon
Hey! Glad to hear you're excited about it! Yes, we're currently working on improving our MCP support in general - we'll have more to share soon, but part of that is supporting SSE servers directly
Very cool. Like I said I can make it work with Stdio but I have a SSE MCP proxy I wrote to combine multiple MCP servers (just to make plugging in all my tools to a new client easier to test). That said, I think after looking at the docs that I'll be tempted to move my tools in directly but I probably will keep them behind MCP for portability.
Happy Mastra user here! Strikes the right balance between letting me build with higher level abstractions but providing lower level controls when needed. I looked at a handful of other frameworks before getting started and the clarity & easy of use of Mastra stood out. Nice work.
I don’t really understand agents. I just don’t get why we need to pretend we have multiple personalities, especially when they’re all using the same model.<p>Can anyone please give me a usecase, that couldn’t be solved with a single API call to a modern LLM (capable of multi-step planning/reasoning) and a proper prompt?<p>Or is this really just about building the prompt, and giving the LLM closer guidance by splitting into multiple calls?<p>I’m specifically <i>not</i> asking about function calling.
If you ignore the word "agent" and autocomplete it in your mind to "step", things will make more sense.<p>Here is an example-- I highlight physical books as I read them with a red pen. Sometimes my highlights are underlines, sometimes I bracket relevant text. I also write some comments in the margins.<p>I want to photograph relevant pages and get the highlights and my comments into plain text. If I send an image of a highlighted/commented page to ChatGPT and ask to get everything into plain text, it doesn't work. It's just not smart enough to do it in one prompt. So, you have to do it in steps. First you ask for the comments. Then for underlined highlights. Then for bracketed highlights. Then you merge the output. Empirically, this produces much better results. (This is a really simple example; but imagine you add summarization or something, then the steps feed into each other)<p>As these things get complicated, you start bumping into repeated problems (like understanding what's happening between each step, tweaking prompts, etc.) Having a library with some nice tooling can help with those. It's not especially magical and nothing you couldn't do yourself. But you also could write Datadog or Splunk yourself. It's just convenient not to.<p>The internet decided to call these types of programs agents, which confuses engineers like you (and me) who tend to think concretely. But if you get past that word, and maybe write an example app or something, I promise these things will make sense.
To add some color to this<p>Anthropic does a good job of breaking down some common architecture around using these components [1] (good outline of this if you prefer video [2]).<p>"Agent" is definitely an overloaded term - the best framing of this I've seen is aligns more closely with the Anthropic definition. Specifically, an "agent" is a GenAI system that dynamically identifies the tasks ("steps" from the parent comment) without having to be instructed that those are the steps. There are obvious parallels to the reasoning capabilities that we've seen released in the latest cut of the foundation models.<p>So for example, the "Agent" would first build a plan for how to address the query, dynamically farm out the steps in that plan to other LLM calls, and then evaluate execution for correctness/success.<p>[1] <a href="https://www.anthropic.com/research/building-effective-agents" rel="nofollow">https://www.anthropic.com/research/building-effective-agents</a>
[2] <a href="https://www.youtube.com/watch?v=pGdZ2SnrKFU" rel="nofollow">https://www.youtube.com/watch?v=pGdZ2SnrKFU</a>
<a href="https://aider.chat/2024/09/26/architect.html" rel="nofollow">https://aider.chat/2024/09/26/architect.html</a><p>"Aider now has experimental support for using two models to complete each coding task:<p>An Architect model is asked to describe how to solve the coding problem.<p>An Editor model is given the Architect’s solution and asked to produce specific code editing instructions to apply those changes to existing source files.<p>Splitting up “code reasoning” and “code editing” in this manner has produced SOTA results on aider’s code editing benchmark. Using o1-preview as the Architect with either DeepSeek or o1-mini as the Editor produced the SOTA score of 85%. Using the Architect/Editor approach also significantly improved the benchmark scores of many models, compared to their previous “solo” baseline scores (striped bars)."<p>In particular, recent discord chat suggests that o3m is the most effective architect and Claude Sonnet is the most effective code editor.
I don't get it either. Watching implementations on YouTube etc it primarily it feels like a load of verbiage trying to carve out a sub-industry, but the meat on the bone just seems to be defining discreet units of AI actions that can be chained into workflows that interact with non-ai services.
> defining discreet units of AI actions that can be chained into workflows that interact with non-ai services.<p>You got. But that is the interesting part! To make AI useful, beyond basic content generation in a chat context you need interaction with the outside world. And you may need iterative workflows that can spawn more work based on the output of those interactions. The focus on Agents as <i>personas</i> is a tangent to the core use case. We could just call this stuff "AI Workflow Orchestration" or something ... and it would remain pretty useful!
AI seems to forget more things as the context window grows. Agents keep scope local and focused, so you can get better/faster results, or use models trained on specific tasks.<p>Just like in real life, there's generalists and experts. Depending on your task you might prefer an expert over a generalist, think f.e. brain surgery versus "summarize this text".
I don't work in prompt engineering but my partner does and she tells me numerous need for agents in cases where you want some technology which goes and seeks things on the live web and then comes back and you want to make sense of that found data with the LLM and pre-written prompts where you use that data as variables, and then possibly go back into the web if the task remains unsolved.
One of the key limitations of even state-of-the-art LLMs is that their coherence and usefulness tend to degrade as the context window grows. When tackling complex workflows, such as customer support automation or code review pipelines - breaking the process into smaller, well-defined tasks allows the model to operate with more relevant and focused context at each step, improving reliability.<p>Additionally, in self-hosted environments, using an agent-based approach can be more cost-effective. Simpler or less computationally intensive tasks can be offloaded to smaller models, which not only reduces costs but also improves response times.<p>That being said, this approach is most effective when dealing with structured workflows that can be logically decomposed. In more open-ended tasks, such as "build me an app," the results can be inconsistent unless the task is well-scoped or has extensive precedent (e.g., generating a simple Pong clone). In such cases, additional oversight and iterative refinement are often necessary.
One way to think about it is job orchestration. You end up with some kind of DAG of work to execute. If all the work you are doing is based on context from the initiation of the workflow, then theoretically you could do everything in a single prompt. But more interesting is when there is some kind of real-world interaction, potentially multiple. Such as a websearch, or executing code, calling an API. Then you take action based on the result of then. Which in turn might trigger another decision to take some other action, iteratively, and potentially branching.
Without checking out this particular framework, the word is sometimes overloaded with that meaning (LLM personality), but actually in software engineering in general, "agent" generally means something with its own inner loop and branching logic (agent as in autonomy). It's a neccessary abstraction when you compose multiple workflows together under the same LLM interface, things like which flow to run next, and edge case handling for each of them etc.
Modularity. We could put all code in a single function, it is possible, but we prefer to organize it differently to make it easier to develop and reason about. Agents are similar
Congrats on launching. I've noticed that switching prompts without edits between different LLM providers has degradation on performance. I'm wondering if you guys have noticed how developers do these "translations", I'm wondering since maybe your eval framework might have data for best practices.
By the developers of Gatsby is a minus, not a plus makes me think this is going to be the next abandonware.
Surprised this is comment is not higher. Gatsby was one of the worst technologies I have worked with in my long career of working with various JS libraries and frameworks. Im sure the team is smart and capable, but I would not be advertising their work with Gatsby.
Same experience, I had the exact same thought. Was new to react and had to make a website… big mistake, wasted so many hours untangling the regex and hacks keeping together Gatsby over the next few years until that website was retired.
Gatsby never made sense to me. Weird design decisions I couldn’t find any plausible reason for. As soon as Next.js became capable of doing SSG I convinced my team to abandon Gatsby. Definitely a minus, sorry.
gatsby was one of the first static react frameworks, now you have things like nextjs remix astro etc...
i dont think abandonware is fair, thats just the way software goes
The character Gatsby didn't function very well either (as far as being a successful person goes, I quite liked the book and he functioned well as a character) :)<p>However, the Gatsby CMS had a couple of things that were really interesting about it - especially runtime type safety through GraphQL and doing headless WordPress.
I don't want to be that person but there are hundreds of other similar frameworks doing more or less the same thing. Do you know why? Because writing a framework that orchestrates a number of tools with a model is the easy part. In fact, most of the time you don't even need a framework. All of these framework focus on the trivial and you can tell that simply by browsing the examples section.<p>This is like 5% of the work. The developer needs to fill the other 95% which involves a lot more things that are strictly outside of scope of the framework.
Some people don't like frameworks. Some people do. We have a little bit of experience building frameworks, so we figured we'd build a good one.
Couldn't agree more. This also looks mostly like a Typescript "port" of Langgraph, and I say "port" because Langgraph has a TS framework already.
True. That's the reason I see a lot of people dropping similar frameworks like LangChain recently: <a href="https://medium.com/thoughts-on-machine-learning/drop-langchain-and-instructor-use-this-alternative-for-structured-output-generation-30f0b503c6d0" rel="nofollow">https://medium.com/thoughts-on-machine-learning/drop-langcha...</a>
i was using vercel ai sdk for my production app and it was such a bad experience that I eventually went with native implementation and tbh it was not much of work thanks to cursor.
problems i faced: too many bugs (just browse their github repo to get an idea), the UI side also had suboptimal performance based on how they implemented hooks.
vercel's whole shtick is to make money off of dumb js devs who do not know better. i think they pay far too much attention on how things look compared to how things work. but hey, they made millions, possibly billions off of those js devs so who is to blame them.
I agree, and it feels like JS is just the wrong runtime for agents. Really languages that can model state in sane ways and have a good concurrency story like Elixir make much more sense.<p>And here’s a fun exercise: ask Claude via Cursor or Perplexity with R1 to create a basic agentic framework for you in your language of choice on top of Instructor.
> good concurrency story like Elixir make much more sense<p>Agree, that's why I've been building this: <a href="https://github.com/agentjido/jido">https://github.com/agentjido/jido</a>
<p><pre><code> > Really languages that can model state in sane ways and have a good concurrency story like Elixir make much more sense.
</code></pre>
Can you expand on this? Curious why JS state modelling falls short here and what's wrong with the concurrency model in JS for agents.
You could describe all frontend JS frameworks the same way: you spend 95% of time on content and mechanics of your webapp, while the framework provides the easy 5%.
Impressive. Have you seen any success with Mastra being used to build voice agents? Our company has been experimenting with VAPI, which just launched a workflow builder into open beta (<a href="https://docs.vapi.ai/workflows">https://docs.vapi.ai/workflows</a>), but it has a lot of rough edges.
We're just starting to do that and have a few TTS providers: ElevenLabs, OpenAI, PlayAI.<p>We hear a lot from people who are outgrowing the voice agent platforms and moving to something like pipecat (in Python), and we'd love to be the JS option.
Is any of the voice stuff in any way 'natural' sounding, I'd love to be able to recreate the ChatGPT app voice experience in my own app with a custom agent, but it just sounds robotic and crap
If you'd like, feel free to reach out to me via email with your requirements and we can get a conversation going. I've built a few voice agent systems in both python and JavaScript and would love to hear about what issues you're running into. Might be able to build what you need.
I basically learned everything about how agents work by using Mastra's framework and going through their documentation. The founders are also super hands-on and love to help!
Kudos on using XState!
Congrats! Side question - is the website OS as well? I'd like to "borrow" the Nav Bar code. I looked on GitHub and couldn't find it in the repos and 300+ branches. Cheers!
This looks really great! How do you make money? Do you charge for deploying these to your platform? I couldnt find anything on pricing
Congrats on launching! Curious how early the Mastra team thinks people should be thinking about evals and setting up a pipeline for them.
We tend to recommend folks spend a few hours writing evals after they spend a couple weeks prototyping. Then they get a sense of how valuable evals are for their use-case.<p>We think about evals a bit like perf monitoring -- it's good to have RUM but also good to have some synthetic stuff in your CI. So if you do find them valuable, useful to do both.
Are there any plans to add automatic retries for individual steps (with configurable max attempts and backoff strategy)?
I created a similar library for orchestrations, but it’s more explicit and lightweight. <a href="https://github.com/langtail/ai-orchestra">https://github.com/langtail/ai-orchestra</a>
Congrats, looks promising!
1. Is it possible to create custom endpoints? I see that several endpoints are created when running “mastra dev”.<p>2. Related to previous question, since this is node based, is it possible to support websockets?
Does Mastra support libraries of tools for agents like toolhouse.ai or <a href="https://github.com/transitive-bullshit/agentic">https://github.com/transitive-bullshit/agentic</a>
Why is it on top of Vercel’s platform?
It looks like theyre using the vercel ai sdk, which really isnt the vercel platform, doesnt have anything to do with any of the rest of vercel. Its actually quite nice and full featured.
It’s not. It’s on top of AI SDK, which is a popular open source library maintained by Vercel.
i am very long on TS as the future of agent applications. nice work team
This looks really nice. We've been considering developing something very similar in-house. Are you guys looking at supporting MLC Web LLM, or someother local models?
Do the workflows support voice-to-voice models like openai's realtime? Or if something like that exists I'd be curios.
Congrats! This is exactly what the AI world needs. I'm thinking about using Mastra for a class I'm working on with AI Agents.
So an AI Mastra Class?
that's awesome!
Interested to learn more about the PDF -> CAD project built on mastra, can you share a link?
I thought Kyle Matthews was the creator of Gatsby
Kyle started the project, I started helping pretty shortly thereafter, then he and I cofounded the company together. Kyle's working on ElectricSQL now but is using us, we're doing a meetup together next month, etc.
I put the "creators" bit in the title because I thought readers would find it interesting. Sorry if that was not-quite-right! I've turned them into developers now.
Got excited, was hoping to see a repository of Go Agents.
Neat, I’m going to use this
Super excited to try out the new agent memory features
interesting to contrast the recent memory releases<p>- <a href="https://mastra.ai/docs/agents/01-agent-memory">https://mastra.ai/docs/agents/01-agent-memory</a><p>- <a href="https://blog.langchain.dev/langmem-sdk-launch/" rel="nofollow">https://blog.langchain.dev/langmem-sdk-launch/</a><p>- <a href="https://help.getzep.com/concepts#adding-memory">https://help.getzep.com/concepts#adding-memory</a><p>not sure where all this is leading yet but glad people are exploring.
100% and agree with this, we saw the langmem stuff last night<p>imho getting some sort of hierarchical memory is conceptually fairly straightforward, the tricky part is having the storage and vector db pieces well integrated so that the apis are clean
let us know what you think!
Congrats guys! really excited to try this out!
Any timeline for python?
You probably already know, but in case you don't, python has phidata[1]<p>[1]: <a href="https://docs.phidata.com/introduction" rel="nofollow">https://docs.phidata.com/introduction</a>
Not planning on it — we think frameworks should be single-language
"You may not provide the software to third parties as a hosted or managed service" - The Elastic v2 license isn't actually open source like your title mentions: "Open-source JS agent framework"<p><a href="https://github.com/mastra-ai/mastra/blob/main/LICENSE">https://github.com/mastra-ai/mastra/blob/main/LICENSE</a>
> Mastra uses the Vercel AI SDK<p>It started off wrong.
Bamfs
Very interesting set of abstractions that address lots of the pain points when building agents, also the team is super eager to help out!
[dead]
You’re awesome guys!
I had so many problems with lanchain and am very happy since switching to Mastra
A TypeScript first AI framework is something that has been missing. How do you work with AI SDK?
idk man<p><a href="https://js.langchain.com/docs/introduction/" rel="nofollow">https://js.langchain.com/docs/introduction/</a><p><a href="https://www.vellum.ai/products/workflows-sdk">https://www.vellum.ai/products/workflows-sdk</a><p><a href="https://github.com/transitive-bullshit/agentic">https://github.com/transitive-bullshit/agentic</a><p>which is not to say any of them got it right or wrong, but it is by no means "missing". the big question w all of them is do they deliver enough value to last. kudos to those who at least try, of course
We originally were wrapping AI SDK, but that confused people who wanted to use both, so we decided to make the API more explicit, eg:<p>import { Agent } from "@mastra/core/agent";
import { openai } from "@ai-sdk/openai";<p>export const myAgent = new Agent({
name: "My Agent",
instructions: "You are a helpful assistant.",
model: openai("gpt-4o-mini"),
});
<a href="https://typedai.dev" rel="nofollow">https://typedai.dev</a> is another full-featured one I've built, with a web UI, multi-user support, code editing agents, CodeAct autonomous agent
Mine is written in TypeScript and I still think it's more ergonomic than anything else I'm seeing in the wild. Maybe there's finally an appetite for this stuff and I should release it. The Mastra dashboard looks pretty nice, might take some notes from it.