Scans and mixed formats.
Files arrive as scans, spreadsheets and email attachments. An agent that cannot read the page cannot cite it.
Structured data for AI · inputs for your Sovereign Artificial Intelligence (AI)
Structured data for AI: getting your files ready for AI agents and tutors
BiltIQ structured data for AI is the data-preparation service behind Agent OS and Learning OS. BiltIQ engineers turn a buyer's documents and records into inputs the products can use: loaded files and written checklists for Agent OS, connectors over MCP, and the assigned course books for the AI Teacher. It is scoped and priced per project.
We'll walk you through it on sample data.
- Formats loaded for Agent OS (by design), scans included
- 30+
- Connectors to your systems, built per deployment
- MCP
- File preparation, inside the Agent OS paid pilot
- Included
- Larger work quoted in writing after the scoping call
- Per scope
Why data comes first
Your AI can only cite what it can find
Rules that live in people's heads.
A checklist that exists only in a senior reviewer's memory cannot be tested. We write it down with you, as agent checks.
Systems that don't talk.
Agents that need a status from your ERP need a connector, not a copy of the database.
What we prepare
What we prepare, for which product
Agent OS
Your past files loaded (30+ formats by design, scans included); your checklist and rules written as agent checks; connectors to your systems over MCP, built per deployment.
Learning OS
The course books your institution assigns, loaded so the AI Teacher can cite chapter and page; the student list and coordinator for the trial from Apr 2027 (targeted).
Business OS
Opening balances set at go-live; inventory, production BOMs and HR attendance imported from Excel/CSV. A direct Tally or Zoho importer is on the roadmap.
Where the data sits
Where your data sits while we prepare it
Agent OS, Own it or pilot
On your GPU server or a BiltIQ loaner, on site. Outbound calls are blocked by default (platform gate). If you have no server, the pilot can run on our India cluster under NDA.
Learning OS
Not installed on your premises today. LMS course files sit in AWS cloud storage (India region). LMS quizzes use a local open model first, with a cloud fallback (Kimi) that is on by default and can be switched off. AI Teacher embeddings, vision and voice default to cloud services and can be repointed.
Business OS
Database on your network on an Own-it install; files and attachments in GCS or S3 cloud storage today. On-prem file storage is on our roadmap. No Claude or OpenAI calls; the Kimi cloud-model option is off by default.
How a project runs
How a data-preparation project runs
Scoping call.
We look at a sample of your files with you and agree what the product needs.
Inside the paid pilot.
For Agent OS, preparing your past files is part of the 4–6 week pilot.
Findings.
The pilot report lists what your AI caught, what it missed and which data problems caused misses.
Ongoing.
Under the managed service, agent tuning covers new checklists as your rules change.
How it is priced
How data preparation is priced
Quoted per scope
There is no list price for preparing data for your AI. Larger work is quoted per scope, in writing, after the scoping call.
Agent OS: included in the paid pilot
₹5–8 LPaid pilot, preparing your files includedex-GST, credited against the licence
FAQ
Frequently asked questions
What does AI-ready data mean?
AI-ready data is information your AI can find, read and act on reliably: files it can load and cite, rules written down as checks it can test, and connectors to the systems it must query. For BiltIQ, that means inputs for Agent OS and the AI Teacher in Learning OS, prepared by BiltIQ engineers and tested against your own manual results.
Do we need to clean our data before an AI pilot?
No. For an Agent OS pilot, BiltIQ loads your past files as they are, scans included, and writes your checklist into the agents. The findings report then shows what your AI caught, what it missed and which data problems caused the misses. You fix what matters before the licence decision, not before the first call.
Does our data leave our premises during data preparation?
Agent OS pilot documents stay on your server, a BiltIQ loaner or, under NDA, BiltIQ's India cluster. Learning OS cannot be installed on-premises today: LMS files sit in AWS (India region), its quiz Kimi fallback is on by default (can be switched off), and AI Teacher embeddings, vision and voice default to cloud (can be repointed). Business OS keeps files in GCS or S3, with Kimi off by default.
What is an MCP connector for AI agents?
MCP, the Model Context Protocol, is an open standard for letting software agents, including your AI agents, call tools and read data in other systems. In BiltIQ Agent OS, the Connect feature speaks MCP, and BiltIQ engineers build each connector per deployment, so your AI agents can query your ERP or document store without a copy of the whole database.
Get your files ready for your AI
We'll walk you through it on sample data.