Kirha Monthly Recap — February 2026

It’s no secret that Kirha’s future is horizontal. We’re building a marketplace of curated data sources exposed through our context graph that AI agents can query on demand across any vertical. Every month we expand into new verticals like Adtech and Developers while going deeper in existing ones.

Being horizontal comes with a real GTM challenge: you can’t speak to everyone at the same time. We needed to figure out which vertical would bring us our first six-figure contracts. In February, we found it. The Medical vertical. In this second month in SF, a lot has happened across tech, hiring, and development. Read on.

Business & Ecosystem

Kirha will power the AI roadmap of Cureety

Kirha is honoured to partner with Cureety to power the next generation of AI in oncology remote patient monitoring.

In this first use case, Kirha’s Search will be integrated directly into Cureety to support clinicians with context-aware medical intelligence. It allows care teams to easily and quickly retrieve trusted, guideline-aligned information, such as treatment safety and drug dosage protocols.

Why does this matter? Oncology is one of the fastest-evolving fields in medicine. Treatment regimens are becoming increasingly complex, while medical knowledge continues to expand at a pace that makes it almost impossible for any individual to absorb. In parallel, many institutions are experimenting with general-purpose AI tools that lack clinical specificity and accuracy. Healthcare does not need more generic AI. It needs contextualized, trusted, and transparent AI.

This partnership reflects a shared ambition: to deploy AI responsibly, transparently, and in direct service of patient care.

About Cureety: Cureety is a CE-marked European oncology-specialized digital medical device solution that provides proactive remote patient monitoring and automated workflows for care teams to improve patient outcomes and quality of life. Today, Cureety supports more than 60,000 patients in collaboration with over 200 healthcare institutions.

ViVE LA vie

Vaughn wanted us to attend the ViVE healthcare conference in Los Angeles. Outbound is cool (or is it?), but meeting professionals where they are is better. I’m usually not a big fan of conferences, but this one hit different.

The standout moment was watching Heidi launch their Evidence solution from a booth that put the rest of us to shame. Evidence is aiming for similar goals as what Kirha will provide for Cureety. What sets us apart is our ability to offer a seamless white-label integration into Cureety’s platform, and that Cureety can run our routing models locally. This means no prompt or patient data ever leaves their environment or goes to Kirha, ensuring privacy is always preserved. In spaces where privacy is paramount, local AI is the only way.

Beyond Heidi, we talked with dozens of people across the space. From an AI medical certification startup based in the EU (hello Sacha @ Scarlet), to the market leader in unified governance for data and AI (cc Ben @ Collibra). We learned a ton. Worth every hour.

San Francisco GTM Lab

I had the chance to meet Vaughn only a couple of days after I arrived in San Francisco in January. The timing was ideal. Vaughn spent 8 years doing sales at LinkedIn and was looking for a new challenge, something far more entrepreneurial and early-stage.

February was the first month we officially worked together, and he’s already taking our sales conversations to another level. And do not get me started on founder-led sales. I hear you, and I am not delegating it. I’m learning from him. A lot. Together, we’ll open the Kirha GTM lab in SF.

What We Shipped

Faster planning powered by Cerebras

You can now choose the runtime environment for the Kirha planner. The fastest option is Cerebras, and it’s now my default. Accuracy remains high. The right tools are still chosen, even for advanced compositions. Latency is now 6x lower.

It’s another plannert. I’ll let Cerebras explain why speed wins and why SRAM eats GPUs for breakfast.

Boost your agents with the Kirha Skill

Every node of Kirha’s context graph is an MCP server, and we have good reasons to keep it that way. But we’ve now built a Skill that tells your agents how to search the graph and it changes the experience entirely.

Skills and MCP are complementary. As Manus says, “Skills provide the playbook; MCP provides the data pipeline”.

Our Skill instructs your agent on how to discover new verticals and providers, so you can just prompt without knowing which part of the context graph you’re searching. It’s probably the easiest way to unleash Claude to its full potential. Soon on skillsmp.com and skills.sh. Until then, read the installation instructions on the Github repository.

Take back control of your search with Kirha SDKs

• Get started with Kirha
  npm install kirha $> Seamlessly integrate Kirha into your workflow and unlock powerful, real-time context for your AI agents. Read the docs.

• Keep your searches private
  If you run local AIs but still rely on Perplexity, Exa, or Linkup for every web search, you’re not really achieving anything but leaking your prompts to them. That’s not happening when you run Kirha’s Local Planner on your machine. Our new planner-sdk-typescript makes it easy to interact with the planner locally. It’s fully compatible with any OpenAI Chat Completions endpoint.

• Supercharge Vercel-based apps
  Use the AI SDK for Vercel to natively integrate Kirha-powered search into your Vercel deployments with minimal configuration.

New Developers vertical

This is a new vertical for AI agents looking for developer context. Which model to download from Hugging Face? Inspect the best training datasets, stay tuned on the latest research papers, download packages safely after scanning for vulnerabilities, search repositories and contributors on GitHub…

Everything your agents need to know to stay safe and up to date with the dev community.

Closing Thoughts

If the future of Kirha is horizontal, we face fierce competition in every vertical on our enterprise GTM. That competition will fade as the models get better. As models improve, vertical tools get commoditized. What remains are the models and the data. That’s exactly where Kirha sits.