We build AI teams for the work
people can only ever spot-check.
YantrAI Labs builds and deploys AI agents inside the systems companies already run. OHM is our finance team. Before it, we put the same idea on shop floors, store counters and government offices.
Vision
Organisations where nothing important goes unchecked.Not more dashboards, and not more people. The check itself becomes something that runs on everything, every time, at the moment it can still change the outcome.
Mission
Put a full-time AI team on the work people can only sample — every invoice, every shift, every counter — so control stops being a matter of how much time someone had that week.
We do that inside the systems already in place. No migration, no rip-and-replace, and no asking a finance or operations team to work somewhere new.
What we believe
Sampling was a compromise, not a standard
Checking a share of transactions was never the goal — it was what a human-sized team could manage. When the constraint lifts, the standard should move with it.
The system of record stays where it is
Your ERP holds the authoritative books. Agents read from it and post back to it. A product that needs to become your ledger is asking for a migration, not offering help.
Access is decided before the AI acts
An agent inherits the signed-in user's scope and cannot exceed it. That ordering is why an agent can be trusted with a vendor master at all.
Throughput, not freed hours
The honest measure is the same team handling several times the volume at the same standard — not a claim about hours saved that nobody can find on a P&L.
Compliance is local
GST filing status, TDS sections, MSME clocks, PAN linkage. Generic finance AI does not know these exist. Most of the recoverable money in an Indian company sits behind exactly them.
Say what happens when it isn't sure
Anything that runs on every transaction will meet cases it cannot settle. A control earns trust by turning that into a named question, not by claiming it never happens.
Where the idea has already run
OHM is not the first time we have put agents on work that was previously sampled. It is the first time we have pointed them at money going out.
| Where | What runs |
|---|---|
| BlueStone | Vision AI across jewellery stores — staff, customer and store analytics on every hour of footage rather than a reviewed sample |
| Government of Odisha | Broker detection and visitor analytics at Sub-Registrar offices |
| Omaxe | Construction and operations analytics across sites |
| Mount Meru Group | Operations across a multi-country group |
| Baxy | Manufacturing operations |
Live in India, Dubai, the USA and Africa.
Who we are
Two founders. Engineers and operators who ran large-scale operations before building software for them.
Mohit
Strategy · Business · AI
B.Tech, IIT Delhi '12. Hero Group, $200M+. Jangid Motors, $5M ARR. Oye! Rickshaw — 100k+ transactions a day, $12M raised from Matrix and Xiaomi.
Rohit
Business · GTM · Product
B.Tech, NIT Rourkela '12. Hero Group. EasyLokal — ~$100K through Techstars, acquired by WayCool. Zomato — 1.5M transactions a month.
Every deployment is led by a founder, without handoffs. At our size that is not a service promise — it is the only arrangement in which the person who hears the problem is the person who changes the product.
The fastest way to understand what we do is to let it run on your data.
Ninety days of invoices, one export, findings back within a working day.
$Check your savings→