Xero Labs

AI that runs on hardware you own.

Xero Labs builds private AI systems — open models, knowledge search and automation agents — that run inside your walls, on infrastructure you control.

Book a call See ForeSight
  • 120Bparameter models, served on our own hardware
  • 37,000+knowledge chunks indexed for support search
  • 40+scheduled jobs run by 7 automation agents
  • 20+ yearsdelivering enterprise platforms
In development

ForeSight

Your metabolic health in one private place. ForeSight brings your CGM readings, lab reports, blood pressure, weight and medications together, then explains in plain language how they connect.

  • Everything in one place. CGM readings from any brand, lab reports from any clinic (PDF or scan), blood pressure, weight and medications.
  • AI that connects the dots. Weekly summaries of what changed and how your numbers relate, plus a report ready for your next doctor visit.
  • Private by design. Your documents, notes and personal details are encrypted with a key unique to you, our staff tools can’t read your health data, and our AI runs on our own private servers.
Join the waitlist

ForeSight is an informational tool, not medical advice. It does not diagnose, recommend doses or change treatment. Always consult your healthcare provider.

Try it: one weekSample data · fictional person
6 am10 am2 pm6 pm10 pm
Daily summary

Help shape ForeSight

Join the waitlist and tell us what you need. Two minutes, seven questions. Early members get free beta access.

Which apply to you? optional
Do you use a CGM?
What do you use today?
What would you want in one place?
Would you pay for it?

Early access

Assay

Automated testing for security platforms. Assay runs full test cycles across every site, explains failures in plain language, and hands you an evidence report ready for sign-off.

  • Every site, one run. The same test cycle across lab, staging and customer sites, over VPN where needed.
  • Deep checks, not just pings. Certificate enrollment, HSM key operations, signing, failover and failback, then a clean slate.
  • Failures explained. A private AI model reads the logs and tells you what broke and what to try first.
  • Evidence you can hand over. Timestamped results and logs in one report for change control and customer acceptance.
Request early access
Test cycle · Release validationExample run against a sample device-identity platform
CheckLabStagingSite 3

Demo with simulated results. No customer systems are involved.

What we build for businesses

Every engagement runs on infrastructure you own or choose. Nothing leaves your control unless you decide it should.

Private AI platforms

Open models up to 120B parameters, deployed on your own servers with load balancing, monitoring and secrets management built in.

gpt-oss 120B · Qwen3-Coder-Next · Qwen3.6 · Qwen3.8 · nomic-embed-text · GPU inference

Support knowledge search

Your wiki, tickets and case history turned into a search your support team can ask questions of, with sources cited and a daily refresh.

RAG · vector search · wikis · tickets · support cases

Automation agents

Agents that run routine checks, reports and triage on a schedule, and ask for approval before anything that matters.

Health checks · security audits · daily briefs

PKI, HSM & IoT security

Architecture, deployment, upgrades and escalation support for device identity and certificate platforms, from people who have run them in production.

PKI · HSM · Linux · public & private cloud · containers

How an engagement works

Four steps, each with something you can see and keep.

  1. Assess

    We map your data, systems and the questions your team asks most. You get a short written plan and a cost estimate.

  2. Design

    Hardware sizing, model choice, security and access rules, agreed with your IT and security teams.

  3. Deploy

    We build it on your hardware or private cloud and prove it against your real questions and test cases.

  4. Operate & hand over

    Monitoring, runbooks and training, so your team can run it — with us on call if you want us.

“The best AI is the kind your operations team can actually run, at 2 a.m., on hardware they own.”

Rajesh Dubey, Founder

Xero Labs was founded by Rajesh Dubey, an enterprise architect with more than two decades in professional services, customer success and security platforms — most recently eight years delivering IoT identity and PKI systems for global enterprises.

The lab runs everything it offers: a multi-node AI cluster, a support knowledge base, a fleet of automation agents and the monitoring around them. What we deploy for you, we have already operated ourselves.

Rajesh on LinkedIn

Tell us what you want AI to do.

[email protected]

A 30-minute call is enough to tell you whether private AI fits your problem, and roughly what it would take.

Book a call