About Chokmah
An AI adoption practice, not a training company.
Training is our delivery mechanism. Workflow throughput is our product. The most persuasive things on this page are the things we refuse to sell, because no other vendor would publish them.
Chokmah is an AI adoption practice in Bengaluru that installs agentic workflows in global capability centres and measures throughput, not logins. A new practice with zero client engagements, it publishes what it refuses to sell and four non-negotiables, because the positioning is the only honest vendor in the room.
- Chokmah refuses to sell anything a course catalogue could deliver, Prompt Engineering 101, or a cohort as a first sale.
- It will not take an engagement where the workflow cannot be named before the work starts.
- Non-negotiable #1: every engagement names the workflows it would refuse to automate, and why.
- Commercial terms are fixed: 50% upfront, milestone-billed, 45-day payment terms in the MSA.
- It sells to the transformation owner, never the L&D head, and has zero client engagements so far.
What Chokmah is
Chokmah is an AI adoption practice in Bengaluru, Karnataka. We install working agentic workflows inside mid-market and nano global capability centres and measure workflow throughput, not licence logins.
We are new, and we say so. Chokmah has shipped zero client engagements. The scenarios on this site describe how we would run the work (composite workflows, real method), and every figure in them is an illustrative target, not a measured result. A practice whose entire positioning is "the only honest vendor in the room" cannot afford a fabricated case study, so there are none here.
The operation is deliberately small: one operator plus contract trainers, cash-positive from a few engagements rather than from a funding round. That shape is a feature. It is why we can afford to tell a prospect which workflows to leave alone, and why we can walk away from a logo that would break the positioning.
Why the honesty is the product
Enterprises bought AI tools and got nothing back. The buyer we want has usually already been sold a hype deck that failed. So we lead with the counter-evidence (MIT's 95%, Gartner's 40%, METR's 19%) each sourced, dated and caveated. If a statistic on this site cannot be sourced, it is cut. If a number is contested, we say so. That is the whole method, applied to ourselves first.
What we refuse to sell
The list no other vendor publishes. Each of these is a sale we will turn down.
Anything nameable in a course catalogue
If it can be delivered asynchronously to a million people, it is already free. We do not resell what the free tier already covers.
Prompt Engineering 101
Free national-scale programmes teach prompt basics, and increasingly orchestration and MCP too. Literacy is free, so charging for it is a mistake. We are paid to apply it to your workflow, not to re-teach it.
Cohorts as a first sale
A cohort is rung 3. Selling it cold, before a diagnostic and a shipped workflow, is how training spend produces certificates and no change.
Engagements where we cannot name the workflow
If we cannot name the workflow before we start, we are selling the thing that fails 95% of the time. We decline.
Logo-chasing discounts
We will not break the pricing or the positioning to win a name for the wall. There is no wall.
Four non-negotiables
The commitments that hold on every engagement, written here so you can hold us to them.
- 1One
We tell you what not to automate
Every engagement names the workflows we would refuse to automate, and why. If we cannot do that, we are selling the thing that fails 95% of the time. Non-negotiable #1 is built into the structure of every scenario page.
- 2Two
Clear commercial terms
50% upfront, milestone-billed, 45-day payment terms written into the MSA. No surprise scope, no open-ended retainer dressed up as a project.
- 3Three
A written record with numbers
Every engagement ends with a written account carrying real numbers, and we ask for two named introductions in the room. We report the number even when it is not flattering.
- 4Four
We sell to the transformation owner
If we are pitching the L&D head, we are in the wrong meeting. The budget is transformation, never L&D, and that decides the outcome as much as the work does.
You have the licences. You don't have the workflows. We install both, and we measure the workflow, not the logins.
See whether we are the right kind of small
A small practice wins when the deliverable is working code in one named workflow and you need the team in the room. Start with a free AI Reality Check and find out.