Loading today’s stories
Leadership1 distinct publisher2 min readPublished
Rishi Katdare of AWS argues that approval should turn on the queue that closes and the report that disappears, because a faster layer above surviving process leaves the company funding the same activity twice.
The Board Room · Leadership desk
Compiled by The Board RoomSomething wrong?How this is made
Follow any of these and your For You feed starts watching them — no settings page required.
build
The nightly shutdown Lambda earns its postmortem on the morning restart1 distinct publisher
security
ONCD stakes Texas water security on six months of donated vendor red teaming1 distinct publisher
build
Partial batch failures: the handler is fine, the mapping decides1 distinct publisher
leadership
Dropping Item 407(j) left boards learning cyber risk from the people they supervise1 distinct publisher
The sequencing matters more than the format. Katdare describes watching reports outlive better data because the report had become part of a management ritual, with the meeting surviving even after the dashboard improved [11]. Once a system is live, the surviving process has its most persuasive defence available: it is the fallback while the new layer earns trust. A removal condition agreed before funding works as a pre-commitment, while the same condition raised after go-live becomes a negotiation with whoever owns the queue, conducted from the weaker side of the table.
The ledger has two sides. Most business cases fill in only one. Katdare's own framing requires the list to account for both the work removed and the new work created around the system [14], which makes net removal the difference between those two quantities rather than the size of the first [1]. A case that presents only retired tasks is offering a gross figure and calling it net, and the gap is not visible on the slide because nobody was asked to size it.
Katdare grants that the idea itself is not new; his claim is that what the list must prove has changed [6]. The procurement review he cites is what gives that claim its edge. In the proof of concept, the recommendation layer was not the hard part, and most effort sat in data cleaning, business interpretation and agreement on the business question being answered [3], against inconsistent product descriptions, unclear units of measure and missing definitions that needed subject matter experts to interpret [4]. The work a demo removes and the work that consumes the budget were not the same work.
The board-deck version of this is an ROI model with hours saved per task and an adoption curve beside it. It is incomplete on Katdare's reading because usage data establishes only that people are using the system [16], which is a fact about behaviour rather than about cost. What the record does not tell us is how often a stop-doing list converts into savings a finance function can book; this is one practitioner's account rather than a study, and the honest position is that the hit rate is unknown. The testable part is the item he wants named at approval, the metric that proves the work stayed gone [7]. A funding decision taken this quarter without that metric will be reviewed a year from now against adoption numbers, because adoption will be the only thing anyone defined.
Ranked by verification strength, evidence, and original report placement.
Rishi Katdare, a senior leader in networking and edge for global financial services at Amazon Web Services, argues in a Forbes Tech Council piece that an AI business case needs a stop doing list.
Katdare writes that many AI business cases become unreliable because they describe what the system will do but not what the organization will stop doing once the system works.
Katdare writes that the missing artifact is not another ROI model but a stop doing list, that the idea itself is not new, and that what changes in the AI era is what the list has to prove.
The discipline, per Katdare, forces leaders to identify the meeting that will end, the report that will disappear, the queue that will shrink, the approval that will narrow and the metric that will prove the work stayed gone.
Katdare writes that leaders have to name the work, the owner, the burden, the decision it supports and the condition for removal.
Under the heading that adoption is not the finish line, Katdare writes that usage data can show that people are using the system.
Distinct publishers with included, body-backed reporting in this cluster.
forbes.com
1 article · August 31, 2026
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
One practitioner essay, nothing counted
Every load in this story is carried by a single Forbes contributor column. The procurement pilot at its centre has no named company, no sector detail, no dates and no scale; the surviving reports are recalled, not documented; the agent supervision cost is anticipated, not observed. The reasoning holds together and the removal tests it proposes could in principle be checked — but nobody outside the author has checked any of it, and not one number appears in the piece.
No deployment record supplied
Nobody ships anything in this story. There is no release, no named firm running a stop-doing list, no pilot count, no before-and-after on a queue that actually closed — only illustrations of assistants that might answer faster while the queue stays. Uptake of the practice is unmeasurable from what we have, so we are not going to estimate it.
Deflationary argument, oversized scope
Unusually, the rhetoric runs downhill: the whole point is to talk executives out of booking savings they have not earned, and the closing line — savings are assumed, not realized — is the opposite of a growth pitch. The overreach is in reach, not in volume. One reviewed procurement effort, a few invented vignettes and some remembered meetings are asked to support a general account of how AI business cases behave, and the headline promise that a list is where economic value becomes real is stated with more certainty than anything on the page can carry.
Vendor byline making an inconvenient ask
The byline cuts both ways and it is worth saying so. Katdare names Amazon Web Services in his first line, and AWS sells into exactly the AI programs under discussion; the venue is Forbes' council channel, where the executive writes and the argument reaches the page without a reporter between. Against that pull, the advice makes AI spending harder to approve — telling buyers to prove a queue closed before they claim savings is a peculiar way to sell cloud. Interested platform, unhelpful-to-self message.
Sure of the argument, unsure of the world
Two very different certainties are tangled here. What was argued, by whom, and under whose banner is beyond doubt — the text is explicit and the disclosure is upfront. Whether AI business cases generally fail for the reason given, and whether stop-doing lists change outcomes, is untested by anything in front of us. That split is the honest ceiling for a single unverified practitioner essay.