Skip to content
Chokmah

Point of view

Point of view

Where the counter-evidence lives. MIT's 95%, METR's 19%, Gartner's 40%, stated by us, not discovered later.

Abstract mark for the essay arguing that the unused learning licence is the real competitor to a workflow engagement
Point of view

Your real competitor is the unused licence you already pay for

We are not competing with another training vendor. We compete with the Coursera, Pluralsight and Copilot seats you already bought and no longer use, and a catalogue seat, however good, cannot change a workflow that exists only inside your organisation.

Abstract gradient mark introducing the essay on why most enterprise GenAI pilots return nothing measurable
Point of view

Why 95% of GenAI pilots fail, and what the surviving 5% did differently

The 95% figure is contested and imperfect, and it still describes your pilot. The failure is not model quality. It is that nobody instrumented the workflow the tool was supposed to change, so no result could ever have been measured.

Abstract mark for the flagship essay on which workflows a company should not automate
Point of view

The workflows you should not automate

Naming what to leave alone is the first deliverable, not a disclaimer. A vendor who cannot tell you which workflows to keep human is selling you the thing that fails 95% of the time, and the refusal list is the most useful page we can hand a transformation owner.

Abstract mark for the essay on the gap between valuing AI governance and having actually built it
Point of view

The governance gap: 83% of Indian executives call it essential, 4% have built it

The AI governance gap is not a disagreement about whether governance matters: almost everyone agrees it does. It is a 79-point gap between saying so and having built it, and that gap is exactly where Gartner's cancelled projects come from.

Abstract mark for the essay on measuring workflow absorption rather than licence logins
Point of view

Measure the workflow, not the logins

Licence logins, completions and prompt counts measure that a tool was opened, not that the work changed. The 95% failure rate is a measurement failure as much as an execution one: organisations counted adoption and never instrumented absorption.

Abstract mark for the essay on the METR trial showing AI slowed experienced developers while they felt faster
Point of view

AI made senior developers 19% slower, and they believed it sped them up

The one rigorous randomised trial of AI coding assistants found experienced developers got slower, not faster, while feeling faster. You cannot manage a productivity claim you have not measured, because the people doing the work are not reliable witnesses to their own speed.

Abstract mark for the essay on why free AI courses teach orchestration but cannot apply it to your workflow
Point of view

The free tier now teaches orchestration. It still cannot run your workflow.

Free curricula now teach agent orchestration, tool use, MCP and evaluation: Microsoft ships them open-source. What no population-scale course can do is apply those concepts to your workflow, your data and your governance. That application is the only thing worth paying for.

Abstract mark for the essay on Gartner's forecast that over 40% of agentic AI projects will be cancelled by 2027
Point of view

Gartner says 40% of agentic AI projects will be cancelled. Here is which 40%.

The 40% cancellation rate is not bad luck or immature technology. It is three named, predictable failures (escalating cost, unclear value, absent risk controls) every one of which is decided before the contract is signed, by whether anyone named the workflow first.

Bring the evidence to your team

Book an AI Reality Check