Lokhesh Ujhoodha

#Mauritian

[Longevity @longevitymediaa]

Dr. David Sinclair says a 2018 study of over 120,000 people found no upper limit to how much fitness protects you from death.

On his own podcast, he broke down 7 truths about how much exercise you actually need:

1) Just 75 minutes a week meaningfully lowers your risk

[Kurrent @kurrent_io]

Claude Code almost deleted my database 💀

The warning signs were there for weeks but I just couldn't see them.

Capacitor watches your coding agent, grades every session on safety, and teaches it to stop before it destroys something.

Free. Supports all major agents. Link below 👇

https://t.co/UILcWzgL3X

[Kurrent @kurrent_io]

My vibe coding bill 10x'd. I fixed it with this free tool.

Capacitor scans your coding agent sessions and answers: "what blew up my bill last month?"

Finds the problem. Fixes it. Your bill goes back to normal.

Free on macOS, Linux & Windows
👉 https://t.co/cfgasBpDBW

[Kurrent @kurrent_io]

Stop repeating yourself to Claude. With Capacitor, your entire team's coding agents get smarter and faster with every session.

👉️ Get Capacitor free: https://t.co/YOs9fDX9Y4

#AgenticAI #CodingAgents #AgenticEngineering #SoftwareFactory

[Lokhesh Ujhoodha @yogisawesome]

@sflorimm for anyone who wants to move from Claude Code to any other coding agent check us out https://t.co/0pcJ0lJQdJ

[Lokhesh Ujhoodha @yogisawesome]

stop handing over compacted sessions from one coding agent to the other, just use https://t.co/QGbx6ISFxi we keep the full session for you for free

[Lokhesh Ujhoodha @yogisawesome]

@benz__002 https://t.co/VbZbi9xQca

[Lokhesh Ujhoodha @yogisawesome]

Shared memory for your team’s coding agents -> https://t.co/VbZbi9xQca

[Kirk Dunn @kirkgdunn]

@satyanadella is pointing to something even bigger than corporate sovereignty in the modern AI era.

The enterprise AI trust boundary isn't just about preventing models from training on your data. It's about ensuring an organization's accumulated intelligence is captured as it happens and also remains its own.

That intelligence isn't just documents or model outputs. It's every event, trace, evaluation, decision, correction, success, failure, and learned pattern generated by humans and AI agents working together. The foundation is an enterprise owned system of record that preserves how work happened, not just what was produced.

That's how human capital and token capital truly compound.

This is the architectural direction we're exploring with @KurrentDB and Capacitor: helping enterprises own the history that becomes their intelligence—not just the software they build.

[Kurrent @kurrent_io]

🤖 Your agent just made up a function that doesn't exist. Then did it again next session. Capacitor evaluates every session, catches the pattern, and feeds the fix into your team's guidance. One catch. Zero repeats.

👉 Try Capacitor free → https://t.co/iEjITum2Ih

#AgenticCoding #CodingAgents #AgenticEngineering #ClaudeCode #Cursor #Codex #Pi

[Kirk Dunn @kirkgdunn]

https://t.co/4hsN4AeazJ

[Lokhesh Ujhoodha @yogisawesome]

Experimental project to verify AI generated code https://t.co/QWgH9tOkiy #verification #vibecoding

[Kurrent @kurrent_io]

Want to build smarter AI-native apps?

This new blog from Lokhesh Ujhoodha shows how to use our #opensource #MCP Server to generate event-native apps, complete with sample code to read/write streams.
https://t.co/z5tPWROHb0

#appdev

[Lokhesh Ujhoodha @yogisawesome]

This is a short timelapse of Claude Desktop using Claude Sonnet 4 to build a 3D Chess Knights only game on top of three.js with move validation and win condition done by projections in KurrentDB. History of moves (legal and illegal) are kept in the game's stream.

[Lokhesh Ujhoodha @yogisawesome]

It's Friday 😄! Let's build a 3D game on top of KurrentDB with our MCP Server. Because keeping state transitions is important for more than just business use cases.

Repo: https://t.co/qAN3nXcqZC
Tutorial: https://t.co/iHRpgSC6YC

[Lokhesh Ujhoodha @yogisawesome]

This makes prototyping not only faster but actually quite fun. And cherry on top is that this is only around 200 lines of code and easily extensible to have your own prompts: https://t.co/iHRpgSC6YC
Repo: https://t.co/KgHBnXMdyY

[Lokhesh Ujhoodha @yogisawesome]

As the saying goes, a picture is worth a thousand prompts 😄. This is a short timelapse of KurrentDB's MCP Server building an application from a picture with projections on KurrentDB.

If you would like to know more: https://t.co/DHAvwe3X36 @kurrent_io @AnthropicAI

[Lokhesh Ujhoodha @yogisawesome]

It will write the projection, then actually proceed to test the code by writing events and then reading the output or state of the projection. If there is any issue during the process (syntax or logical errors) the model knows to debug it and repeat the process.

[Kurrent @kurrent_io]

Microservices are hard enough. Your data layer shouldn’t be.

Kurrent Cloud delivers:

🔹 Zero setup hassle
🔹 Data consistency & integrity
🔹 Flexibility to scale with your future needs

🟢 Get up to $200 in free credits: https://t.co/WC0zDAutKR

#appdev