AI Lab
We research and build
New AI frameworks for video analysis, digital twins, robotics simulation and AI devices.
Explore the AI LabAI lab · Custom AI frameworks · Your own servers
An AI lab that builds custom AI frameworks and products, on your own servers.
BiltIQ AI is an Indian AI lab and custom AI framework company. We build AI frameworks, then turn them into custom systems and ready-to-deploy products that run on your own servers: AI Factory, Agent OS, Business OS and Learning OS. BiltIQ AI is the brand of Aarna Tech Consultants Pvt. Ltd., Jamshedpur.
- Countries with paying clients
- 5
- Engineers, 80+ combined years
- 10
- Paid pilot on your own documents
- 4–6 weeks
- Patent applications filed
- 4
- NVIDIA: Inception Partner
- DPIIT: Recognised startup · DIPP239966
- ISO/IEC 27001:2022: Certified
- ISO 9001:2015: Certified
- PM-JAY AI Challenge 2026: Winner (hackathon)
What we do
One team: an AI lab, custom builds and products
Custom AI
We build it for you
AI built around your process and your data, running on your servers, with our engineers behind it.
See custom AIProducts
Ready to deploy
AI Factory, Agent OS, Business OS and Learning OS: what the lab has already turned into products.
See the productsTwo ways to buy our products
Own your AI, or subscribe to ours
Own it
Subscribe
For
Enterprises leaving SAP or Salesforce, large and medium firms, defence, PSUs and government, large colleges and universities
Small firms and MSMEs, smaller colleges and schools, professionals, students
Where your AI runs
Your own GPU server, inside your walls
BiltIQ's GPU cluster in India
What you pay
One-time site licence + yearly managed service. Hardware at cost.
SaaS subscription + API tokens
Your data
Models run on your GPU server, and outside connections from Agent OS are blocked by default. Business OS stores file attachments in cloud storage for now; on-premise file storage is planned.
Inference runs on our GPU cluster in India.
Status
Available now for AI Factory and Agent OS, starting with a paid pilot.
Subscription: coming in 2027.
Own it
- For
- Enterprises leaving SAP or Salesforce, large and medium firms, defence, PSUs and government, large colleges and universities
- Where your AI runs
- Your own GPU server, inside your walls
- What you pay
- One-time site licence + yearly managed service. Hardware at cost.
- Your data
- Models run on your GPU server, and outside connections from Agent OS are blocked by default. Business OS stores file attachments in cloud storage for now; on-premise file storage is planned.
- Status
- Available now for AI Factory and Agent OS, starting with a paid pilot.
Subscribe
- For
- Small firms and MSMEs, smaller colleges and schools, professionals, students
- Where your AI runs
- BiltIQ's GPU cluster in India
- What you pay
- SaaS subscription + API tokens
- Your data
- Inference runs on our GPU cluster in India.
- Status
- Subscription: coming in 2027.
The platform
One AI platform: AI Factory and Agent OS, with three apps on top
- AppsAgent OS workflows (tenders, contracts, claims)Business OS (ERP + CRM)Learning OS (LMS + AI Teacher)
- Agent OSAI agents, security controls and human approval for important actions.
- AI Factory (the base)runs the AI models. You choose open-weight models, with NVIDIA Nemotron as the default for defence.
- GPUyour own GPU server, or our cluster in India (subscription: coming soon)
Four products
Four products. One sovereign AI platform.
AI Factory
The base · Private AI on your servers
Your GPU server, your AI models and the tools to run them, set up and managed by BiltIQ.
On-premise LLM platformAgent OS
Flagship · AI agents with human approval
Agents read tenders, contracts and claims, check them against your rules and show the source page. A person approves what matters.
AI agents with human approvalBusiness OS
ERP and CRM with AI built in
Accounts, GST, sales, stock and HR on one pipeline. 66 copilot tools; the 23 that change data wait for a person.
AI ERP and CRM with GSTLearning OS
LMS and AI Teacher
An AI Teacher answers from your course book and shows chapter and page, in English, Hindi and Hinglish, alongside a full LMS.
AI tutor that cites the bookWhy sovereign AI
Why on-premise AI, and why now
The rules
Defence and government buyers check each model's origin under Public Procurement No. 1. The DPDP Act sets duties and penalties for how organisations handle personal data.
The regulators
RBI's FREE-AI report recommends a board-approved AI policy. SEBI holds the regulated firm answerable for what its models produce, even through third-party tools.
The cost
Rented cloud models are metered: every new user adds to the bill. Owning your AI costs a one-time licence plus a yearly service.
The work
Tenders, contracts and claims are still checked by hand, page by page.
We compete on finished, controlled workflows on your own hardware, not on the model alone.
Pricing
Prices for your AI, in ranges, before you call
| Offer | Price (ex-GST) | Status |
|---|---|---|
| Own it: site licence | ₹50–90 L one-time + managed service ₹16–18 L/yr; hardware at cost | On sale now |
| Paid pilot | ₹5–8 L, 4–6 weeks, credited against the licence | On sale now |
| Subscribe: API tokens | ₹8 per 1M input / ₹28 per 1M output tokens | Subscription: coming in 2027 |
| AI Tutor (Learning OS) | Colleges ₹999 per semester; schools ₹300–1,000 per student-year | Launch price, from Apr 2027. Pilot in Jamshedpur. |
| Business OS subscription | India ₹3,500/month per firm | Launch pricing; coming soon. Opens Jul 2027 (planned) |
Who builds your AI
12 people, 10 engineers, 80+ combined years. Founded in Jamshedpur in 2021 by Harish Kumar (Co-Founder & CEO, 18 yrs) and Sonali Harish (Co-Founder & Director, product and operations). Bootstrapped on services, with paying clients in five countries.
Industries
One platform, configured for your sector.
- Healthcare and claimsClaim and bill checks against policy, with the page cited
- Finance and bankingContract and document checks on your own servers
- EducationAn LMS and an AI Teacher that answers from your books
- Government, defence and PSUTender and contract scrutiny; Nemotron by default for defence
- ManufacturingERP, BOMs and supplier quotes, with approvals
Latest from the blog
Ideas from the people building the stack.
The five-stage AI transformation roadmapA sequenced path from sizing to ownership — what happens at each stage, what you own when it ends, and the gate that must close before the next begins.
What AI transformation actually means for an Indian enterpriseAI transformation is not buying a model. It is building the layer around one — retrieval, agents, a privacy rail, and a record an auditor accepts.
AI transformation in regulated sectorsHealthcare, BFSI, government and manufacturing face different binding constraints. What actually governs each, and what it means for the architecture.FAQ
Frequently asked questions
What does BiltIQ AI do?
BiltIQ AI is an Indian AI lab and custom AI framework company. Our lab builds new AI frameworks, our engineers build custom AI systems for clients, and we sell ready-to-deploy products: AI Factory, Agent OS, Business OS and Learning OS. Most of it runs on your own servers.
Do you have paying customers?
Yes. Our custom AI and software work has paying clients in five countries. On the product side, AI Factory has a reference install at an engineering college, and our agents won the PM-JAY AI Challenge 2026. Our first product licence is planned for early 2027.
What is sovereign AI?
Sovereign AI means your organisation controls the models, the hardware they run on and the rules they follow. With BiltIQ you can own it on your own GPU server, installed and managed by us, or subscribe to our GPU cluster in India once the subscription opens in 2027.
Can we run an LLM on our own servers in India?
Yes. BiltIQ AI Factory installs on your own GPU server and serves open-weight models there, and BiltIQ runs your AI as a managed service. You choose the model: open-weight models selected for your work, with NVIDIA Nemotron as the default for defence and PSU. Each model runs on its own GPU node.
What does it cost to own BiltIQ on our own servers?
Owning your AI with BiltIQ costs a one-time site licence of ₹50–90 L (₹35 L for a platform-only site), plus a managed service of ₹16–18 L a year. Hardware is passed through at cost. Buyers start with a paid pilot of ₹5–8 L over 4–6 weeks, credited against the licence. All prices are ex-GST.
Does BiltIQ offer a cloud or SaaS option?
Yes, as the Subscribe tier: SaaS and API tokens for your AI, served from BiltIQ's own GPU cluster in India. Tokens are priced at ₹8 per 1M input and ₹28 per 1M output, ex-GST. The subscription is coming soon. Join the 2027 waitlist and BiltIQ will tell you when it opens.
Which models does BiltIQ use?
You choose the model: open-weight models selected for your work, with NVIDIA Nemotron as the default for defence and PSU. Your AI runs them on your own GPU server, or on BiltIQ's cluster in India for subscribers. Fine-tuned domain models are a roadmap item after the next GPU expansion.
Can AI agents change our data without approval?
Not for consequential steps. In BiltIQ Agent OS, outbound calls are blocked by default, and irreversible and high-consequence actions pause for human approval. If no approver is available, the action is blocked. In Business OS, all 23 copilot tools that change data wait for a person before anything is written.
Is BiltIQ ISO 27001 certified?
Yes. BiltIQ AI (Aarna Tech Consultants Pvt. Ltd.) is ISO/IEC 27001:2022 certified, certificate 305026060958IS, valid to 8 June 2029. It is also ISO 9001:2015 certified. BiltIQ is a DPIIT-recognised startup (DIPP239966) and an NVIDIA Inception Partner, based in Jamshedpur, Jharkhand.
How does a paid pilot work?
BiltIQ signs an NDA first. Your AI then runs for 4–6 weeks on your own past documents, on your GPU server or a BiltIQ loaner, so the documents stay on site. You get a findings report against your manual results, then decide on the licence. The pilot costs ₹5–8 L, ex-GST, credited against the licence.
Book a scoping callOr write to [email protected]
Tell us what you need. Talk to an engineer.
A 30-minute call ends with a written next step. We walk you through the work on sample data, never on client documents.