Jaseci Labs

Jaseci is an end-to-end open-source and Open Computational Model, Technology Stack, and Methodology for bleeding edge AI. It enables developers to rapidly build

[Jaseci Labs @Jaseci_Org]

Full write-up, side-by-side Python vs Jac:
https://t.co/Cp1Qc6IzNu

Code: https://t.co/Q7b2J0pagR

I'm a Jaseci contributor, happy to take the hard questions.

[Jaseci Labs @Jaseci_Org]

Most of an agent's code is not the agent.

It's schemas declared twice, ReAct loops written by hand, and workflows buried inside prompt strings.

Here's what building agents looks like in a language that treats them as a first-class feature.

[Jaseci Labs @Jaseci_Org]

And the workflow isn't a prompt string, it's a graph the agent walks: each step a node, a walker carrying state through them.

Add it up and the boilerplate moves into the runtime. What's left is the part that was ever actually the agent.

[Jaseci Labs @Jaseci_Org]

A ~12-line Python LLM call becomes one line in Jac:

def answer(question: str) -> str by llm();

Structured output? The return type is the schema, validated and retried for you. Tool use? Pass the functions, the runtime runs the loop. No JSON schema, no dispatch, no retry code.

[Jaseci Labs @Jaseci_Org]

Every Python/TS agent code repeatedly rewires the same plumbing: duplicate intent, control flow in a system prompt, the same ReAct loop.

What if the language already knew what an agent was? In Jac, the signature IS the prompt, the return type IS the schema, tools are functions.

[Jaseci Labs @Jaseci_Org]

Open source repo: https://t.co/opAxe8cKVr

[Jaseci Labs @Jaseci_Org]

Developing MCP servers is 90% "Why didn't that tool call work?" and 10% actual coding.

Is it a schema hallucination? A bad param? Or did I just break the server?

We built ProtoMCP, so we can see exactly what the model is sending vs. what your server is doing.

[Jaseci Labs @Jaseci_Org]

AI isn’t making everyone smarter.
It’s making some people superhuman — and others worse.
As @jasonmars says: the gap is growing fast.
What separates the top from the average now? 👀

🎥 https://t.co/HnyHnIwoPC
#AI #Coding #SoftwareEngineering #Jaseci

[Jaseci Labs @Jaseci_Org]

🚀 NSF awards Jaseci Labs to expand the open-source Jaseci ecosystem!

🔹 Devs: Build agentic AI apps that scale with our AI-first model

👉 UMich CS press story: https://t.co/KmVNs2A6So

#AI #OpenSource #Jaseci

[Jaseci Labs @Jaseci_Org]

Why it matters for devs:
✅ No manual prompt engineering
✅ Seamless model integration
✅ Build → scale from prototype to production effortlessly

#jaseci #developer #AI #AgenticAI

[Jaseci Labs @Jaseci_Org]

What is Jaseci?

It’s a full stack for building agentic AI applications fast
🔶Jac language → Python-variant with AI-first constructs
🔶 Jaseci runtime → Scale-native engine for serving millions of users

Info: https://t.co/PDKGqPeOjR

#OpenSource #ai #agent #jaseci #Developers

[Jaseci Labs @Jaseci_Org]

🚀 Big news: Jaseci has been awarded a prestigious National Science Foundation (@NSF ) POSE award! 🎉

This recognition supports leading open-source ecosystems shaping the future of technology.

#opensource #AI #developers

[Jaseci Labs @Jaseci_Org]

Had a great time at @PyOhio 2025 connecting with Python devs & OSS enthusiasts! 🚀

We shared how Jaseci, our open-source AI-first language & runtime, helps build agentic AI apps & scale from local to cloud.

#python #AgenticAI #opensource #jaseci #AI #programming

[Jaseci Labs @Jaseci_Org]

@rgjust2000 Object-Spatial Programming takes a different approach by moving computation to the data itself. It’s a more intuitive and efficient way to work with connected systems like graphs, networks, and real-world relationships.
Check out here for more info - https://t.co/BfCe5MqBZi

[Jaseci Labs @Jaseci_Org]

🚀 Why build apps the old way…
when you can build AI-native from day one?

Meet Jaseci → open-source, scalable, and built for the future.
👉 https://t.co/YONAwTHrfA

#AI #ArtificialIntelligence #LLM #AgenticAI #Python #OpenSource #Developers #Innovation #Jaseci