FAQ

An AI adoption diagnostic is a time-boxed assessment that interviews the people who run your workflows, shadows two of them, and audits how your existing AI licences are actually used. It ends in a scored report naming which workflows to automate, which to leave alone, and why. It produces a decision, not a strategy deck. At Chokmah it runs two weeks and costs ₹1.5–3L.

Two weeks. Chokmah interviews 8–12 people, shadows two workflows end to end, and audits your current licence utilisation, then delivers a scored report. Two weeks is long enough to see how the work really happens and short enough that you find out cheaply whether there is a workflow worth building against, before committing to a build.

You are ready when you can name one workflow that is high-frequency, rule-dense and already producing a record you can baseline. You are not ready if the honest answer is 'AI, generally'. The two-week diagnostic exists precisely to answer this question with evidence rather than optimism, and sometimes its most valuable output is 'not yet, and here is what to fix first'.

No. Training is Chokmah's delivery mechanism, not its product. The purchasable outcome is a workflow that runs faster or with fewer errors. Anything deliverable asynchronously to a million people is already free, including, increasingly, orchestration, tool use, MCP and evaluation as open tutorials. What no free programme can do is apply those to your workflow, your data and your governance. That application is what we sell.

Pricing is published and indicative: the AI Reality Check is free; the Adoption Diagnostic is ₹1.5–3L; the Workflow Sprint is ₹6–12L; the Capability Cohort is ₹10–20L; the governance and CoE retainer is ₹2–5L per month. Every engagement is quoted individually. These are opening ranges in a market where no India enterprise GenAI vendor publishes a price.

Because the India enterprise GenAI market is quote-gated: no major vendor publishes a price, so buyers cannot benchmark and everything is priced on value and reference. Chokmah publishes its ranges anyway. In a market where secrecy is the norm, being the one practice that prints its numbers is itself the differentiator, and it disqualifies tyre-kickers early.

Transformation, never L&D. The outcome you are buying is a changed workflow, which is a transformation result, not a training one. GCC headcount and L&D budgets are tightening while transformation spend holds. Funding the work from the wrong line (and pitching the L&D head instead of the transformation owner) is a common reason adoption stalls.

Fixed and written into the Master Services Agreement: 50% upfront, the balance milestone-billed, on 45-day payment terms. Chokmah does not discount to chase a logo, because a discount that breaks the pricing also breaks the positioning. The commercial terms are the same for everyone.

Yes. In a Workflow Sprint your team builds the solution alongside us and owns the code, the evaluation harness and the runbook at handover. An artefact nobody inside your team can change six months later is a liability, not a deliverable, so ownership and a working handover are part of the engagement rather than an add-on or an upsell.

A Workflow Sprint is a four-to-six-week engagement on one named workflow, built jointly with three to five of your own people, who keep the code. It ships a working agentic solution plus an evaluation harness that scores outputs against fixed cases. It is the rung where Chokmah delivers working software rather than a plan, and it follows the diagnostic, never the other way round.

One that is high-frequency, high-friction, rule-dense and already instrumented, so you can prove a delta. MIT found more than half of GenAI budgets went to visible sales and marketing use cases despite better returns in back-office automation (MIT NANDA, 2025). Start where the work is boring, measurable and nobody is watching the demo.

A scored, written report, not a strategy deck. It names three workflows worth automating, ranked, with the reasoning; the workflows to leave alone and why; what your current licences are and are not doing; and a recommended first build with its baseline defined. If the honest finding is that nothing is ready, the report says so, which is more valuable than a build that fails.

The documented policies, roles, controls and records that decide who may deploy an AI system, on what data, with what testing, and who is accountable when it fails, typically a model register, an evaluation harness, human-in-the-loop checkpoints, an escalation path and a review cadence. IBM found only 4% of Indian firms have embedded such frameworks, while 83% call governance essential (IBM IBV, 2025).

Yes, in practice. India governs AI through existing law rather than a standalone statute, and the Digital Personal Data Protection Act, 2023 with its 2025 Rules constrains how personal data is used to train and run AI systems. For a GCC the practical obligations are consent provenance for data, audit trails, and a named person accountable for AI risk, which is exactly what a governance framework provides.

A named person must be, and that is half the problem: in most organisations no one owns it. A governance framework fixes accountability explicitly: who signs off a deployment, who owns the evaluation harness, who is called when an agent misbehaves. Chokmah's governance and CoE retainer stands up that framework and enables an internal champion to run it after we step back.

No. Chokmah is a new practice and has shipped zero client engagements. There are no client names, logos, quotes or claimed outcomes anywhere on the site. The scenario pages describe how an engagement would run using composite workflows, and every figure on them is an illustrative target, not a measured result. We would rather say this plainly than manufacture a case study you could disprove in one call.

For transformation owners at mid-market and nano GCCs in Bangalore and Karnataka, funded Bangalore SaaS teams, and logistics or freight shared services. Not for large Indian IT services firms, which run this in-house at a scale a small practice cannot match, nor Fortune 500 MNC GCCs bound by global master-vendor agreements. Chokmah declines both rather than take a meeting it is the wrong fit for.

We cite them with their caveats. The MIT 95% figure has been contested for its small base; we think it holds directionally and say so. METR has stated its 19%-slower result is historical and may not reflect current tools, so every use of it carries that note. Citing a source honestly, including where it is disputed, is more credible than repeating a headline, and it is the whole point of the positioning.