Who cannot use AI at all. The institutions that hold the most sensitive records are the ones that have no way to use cloud AI. We counted them from public registries.
The usual AI market story counts seats at companies that can already buy a cloud subscription. ANVE is for the offices that cannot. The rule that their records must stay where they are held already exists: the DPDP Act 2023, the MeghRaj data-residency guidance for government, departmental security rules, and the professional duties of confidentiality. The tooling has not caught up with the rule.
This note is the count, so the size of that gap is a number and not a feeling.
01
The rule already exists
The DPDP Act 2023 covers personal data. It sets purpose limitation and breach duties. It applies to hospitals, firms and colleges.
The MeghRaj guidance covers government data. It says that data should sit on government-controlled infrastructure. It applies to ministries, departments, districts and courts.
Professional duties cover advocates and chartered accountants. Client confidentiality is set by statute and by the rules of their councils. It applies to every file the firm holds.
None of these forbid AI. All of them forbid sending the record to a server the institution does not control.
02
Public institutions, counted
Counted, not surveyed. Each row is a public registry and the number it publishes.
Correction, 6 October 2026. Re-checking each source: ICAI now lists 27,426 CA partnership firms (1 April 2026); the Local Government Directory lists 784 districts; eCourts shows 3,716 district and taluka court complexes; SCOPE reports 291 operating CPSEs (289 counts only profit- and loss-making ones); AISHE 2023-24 registers 64,756 higher-education institutions, of which 59,533 responded; the hospital counts are a 2020 modelled estimate, not a census; and we could not re-verify the law-firm count. The per-sector pages under /for use the re-checked figures with their sources.
Registry
Count
What they hold
Districts (Government of India, Dec 2025)
800
Land, welfare, vigilance, RTI files
District courts (Supreme Court e-Committee)
743 district courts across 3,443 court complexes
Case files, evidence
Government hospitals
25,000+
Patient records, public-health data
Central public sector enterprises (Public Enterprises Survey 2024-25)
289 operating CPSEs
Tenders, board papers
Central ministries and departments
58 ministries, 93 departments
Everything
About 27,000 public institutions that can own an appliance and cannot use a cloud model.
03
Private institutions, counted
Same method. Each row is a registry and the number it publishes.
The 1,299 NABH-accredited hospitals are the tier that already runs a compliance function. That is the serious hospital buyer.
04
What "cannot" looks like on the ground
Four patterns. None of them is a named customer.
File notes and minutes written by hand, or typed into a consumer chatbot on a personal phone.
A metro hospital chain buying enterprise AI while the district hospital cannot sign that contract.
An examination office that keeps AI out entirely, because one leak ends careers.
A firm that quietly uses ChatGPT and hopes nobody asks.
The gap is not appetite. It is a purchasable, ownable form factor.
05
What this means for what we build
One owned box on the institution's own network.
Many named users through a browser.
An audit trail the institution keeps.
A price a district hospital or a two-partner firm can approve. (Update, 7 October 2026: the ₹95,000 target is retired; indicative launch prices now start at ₹2,99,000 excluding GST. See the lineup.)
An offline entitlement path for sites with no internet. Not built yet (Track G).
Government sales additionally need BIS/CRO registration and a GeM listing. Those are on the plan, not done. We sell to private institutions first and build the government path in parallel.
06
What is not real yet
No sealed unit exists. Every image of the appliance on this site is a labelled render.
The software runs on a development machine and has been used in one real working setting.
No customers, no pre-orders confirmed. Selling has not started.
No certifications. No delivery date.
Final price not announced.
Reservations are ₹999 in India or US$12 elsewhere, refundable in full until you confirm an order in writing. The terms are two screens long and say so. If you run one of the institutions in sections 02 or 03, we are taking ten design partners.
