DataAgent raises $10M to fix production faults in-cluster

DataAgent came out of stealth today with $10 million in pre-seed funding and a claim that inverts how most monitoring is sold: the product is not the dashboard, it is the fix. Its agents run inside your Kubernetes control plane and remediate faults where they happen, rather than shipping telemetry to a vendor and billing you for the privilege of looking at it.

What actually happened

The round was led by MizMaa Ventures and Alicorn Venture Partners. Founders Ishay Yaari (CEO) and Nati Shalom (CTO) started the company in January; both came from Cloudify, the cloud orchestration company Dell acquired in 2023. The team is 15 people and hiring in the US and Israel (PR Newswire, SiliconANGLE).

The architecture is the argument. Agents deploy as a lightweight overlay alongside your existing observability tools rather than replacing them, capture signals at the source, and act to restore service before deep diagnostics run. Because data is processed in-cluster, it never becomes egress. The company projects that can cut observability spend by up to 90% — that is a vendor projection, not a measured customer result, and we would treat it as one. There is a free open-source agent you can deploy standalone, with a paid SaaS tier for fleet management and orchestration.

Yaari's framing: in DataAgent, observability is one feature, and the real product is self-healing infrastructure.

Why in-cluster remediation matters for your business

Most small operators do not run Kubernetes, and this post is not an argument that you should. The transferable idea is about where the compute happens relative to the data.

The standard observability deal is: ship everything out, pay per gigabyte ingested, pay again per gigabyte retained, then pay a human to read it at 2am. The bill scales with how much you log, which means the rational move becomes logging less — and you end up blind in exactly the places you were nervous about. Moving the analysis to the data flips the incentive. You can afford to keep the noisy signals because you are not paying to move them.

The second idea is worth more than the first: remediate first, diagnose second. Restart the stuck worker, drain the bad node, roll back the deploy — then figure out why while the site is up. Most teams have that ordering backwards, and it is the difference between a four-minute blip and a Slack thread that eats a morning.

Neither of those requires a $10M startup. It requires knowing which three failures actually take you down and writing the automatic response before the next one.

Key takeaways

DataAgent launched with $10M pre-seed led by MizMaa Ventures and Alicorn Venture Partners
Founded in January by Ishay Yaari and Nati Shalom, both previously at Cloudify (acquired by Dell in 2023)
Agents run inside the Kubernetes control plane as an overlay on existing observability tools, not a replacement
Processing in-cluster removes telemetry egress; the company projects up to 90% lower observability spend — a vendor claim, not measured customer data
Free open-source standalone agent, paid SaaS tier for fleet management
The portable lessons: move compute to the data, and remediate before you diagnose

You do not need a self-healing platform. You need the top three failures to fix themselves. We build the runbooks-as-code that turn your worst 2am pages into a job that already ran. See what we automate, or tell us what broke last month.

Sources: PR Newswire, SiliconANGLE.

DataAgent raises $10M to fix production faults in-cluster

What actually happened

Why in-cluster remediation matters for your business

Keep reading

GitHub Copilot seats go prepaid as signups reopen Sept 1

Fake AI crawlers hunt .env files. User-agents aren't auth.