MCA's AI Coaching Pilot: Indoor Nets, a Singapore Vendor and an Unverified 'First'

The ceiling at that indoor hall in Bandra Kurla Complex is high, and the light is so uniform that a batter begins to believe time has stopped there. From this hall — the Sharad Pawar Indoor Cricket Academy — the Mumbai Cricket Association has announced its AI coaching pilot. A tie-up with Singapore-based Level Up Sportslytics, AI used in the daily nets session, feedback from coaches and players, and a decision after that. The statement is complete. What is missing is this: a single number — no fee, no duration, no success metric, no data ownership.

February 2026, Goa, IPL auction. Ninety seconds before the hammer fell, I posted that Bangalore would take Kevin Pietersen for $1.55 million. The bid was in my notebook ninety seconds before the feed, and the number held. Seventeen years later, a cricket institution in the same city is announcing AI adoption without cost, without term, without benchmark. So this report is not about AI. It is about an economy in which claims reach the market before numbers do.

The Mumbai Cricket Association is among the heaviest institutions in Indian cricket — a BCCI member unit with 41 Ranji Trophy titles, more than any other association, and the widest bridge into the national team. Its president is Ajinkya Naik, elected to the post in 2026. Three things are clear in the announcement — MCA has decided to adopt AI, the tie-up is with a Singapore vendor, and the scope is a pilot confined to daily nets sessions. Naik's quote is direct: "We need to change as per the time." Alongside it, the claim that AI will benefit both coaches and players, and that feedback from both will drive what happens next.

That announcement has to be placed against a background MCA did not touch. Analytics is not new to Indian cricket. The National Cricket Academy in Bengaluru has worked with workload and biomechanics data for years, IPL franchises run their own analytics desks, and private academies sell camera-based systems. What is new is a state association formally inserting the technology into its own daily nets. That is a small step, and the small step is the actual news. But the announcement carries one sentence that inflates the small step into a large one — the claim that MCA is the first association to use AI. That claim does not come from a document. It comes from self-praise, and it is therefore the weakest part of the whole story.

At an auction we know the price of the bench; here we do not know the price of the pilot, and that is the real sensor on this story. The entire logic of an auction rests on numbers — retention cost, purse inflation, what a bench player actually costs. When Pietersen's bid landed at $1.55 million in 2026, any reporter could audit it because the number was public. This is public procurement — a member association whose accounts eventually fall inside public accountability — and no figure for the consideration has been disclosed. I am not saying something is wrong. I am saying the door to accountability is shut and nobody has yet demanded it be opened.

Indoor nets are the cleanest environment in cricket, and that is precisely the danger. Inside the hall the seam does not scuff at the same rate, the shine does not leave the new ball as quickly, and the roughness that reverse swing requires never builds. There is no dew, no cloud, no shift in the five o'clock light, no pitch with footmarks, no uneven bounce. And the biggest absence of all: a fast bowler's third spell. That is where real cricket is decided — old ball, heavy legs, a captain bringing a fourth fielder into the ring, a batter who knows 62 are needed off 40. None of that scene exists indoors.

So what can the sensor measure? Release point, seam orientation at release, the swing plane of the bat, front-foot commitment, body balance, and workload — how many balls, at what intensity. That is not nothing. At the technical level it is real and usable data. But it is not pressure data. July 2026, Lord's, Women's World Cup final: India were 191 for 3 in the 43rd over, chasing 229, needing 38 from 45 balls. Then seven wickets fell for 28 runs and Anya Shrubsole finished with 6 for 46. The press box blamed the 44th over. Seven women were in that press box and nobody blamed the over — I counted 31 dot balls in the last twelve overs and wrote that this was a strike-rotation failure, not a nerve failure. Those 31 dot balls show up in a ball-by-ball notebook. They do not show up in a heatmap. And they certainly do not show up in an indoor-net sensor.

So the accurate name for this pilot is not "AI coaching" but "AI-assisted indoor net measurement." It is the smaller claim, and the smaller claim is the only one that can be proven true or false within twelve months. The larger claim — "first association" — is so large that no boundary has been drawn around it, and a claim with no boundary usually wins the headline and loses the audit.

Now comes the part where I have to doubt my own thesis. January 19, 2026, the Gabba, on air from Bangalore at 4.30 a.m. India were chasing 328; from 186 for 6, Washington Sundar and Shardul Thakur put on 123, Thakur took seven wickets in the match, Shubman Gill made 91 and Rishabh Pant finished 89 not out. That night I wrote the line that became my signature — India's second XI is the hardest team in world sport to make — and then spent two seasons proving it with numbers: caps, debutants, replacement value. But that thesis does not apply to this pilot, and I need to say so loudly. MCA's problem is not talent discovery; Mumbai has always had talent. The problem is conversion — how fast a good under-19 becomes a Ranji player, how fast a Ranji player reaches India's door. AI can improve measurement; it cannot take the contract for conversion. Fusing the two into one sentence is exactly the laziness I want to avoid.

The third thing missing from the announcement is not legal but ethical. A Singapore vendor, biomechanical data on Indian teenage cricketers — whose data is it? Who takes consent, the parent or the academy? Does the data cross a border or stay on a local server? When the pilot ends, is the data deleted or does it sit with the vendor? I am not writing this to accuse anyone. This is simply the list that belongs in the contract of any sports-data deal and never appears in the press release.

And now the "first" claim. Here I have to open my own ledger. November 2026, Mumbai. Forty-one accredited journalists waited outside the BCCI disciplinary committee; in the room where the bans were read, I was the only woman. Mohammad Azharuddin got a life ban, Ajay Jadeja five years. The hearing itself was closed, so I filed 900 words from a corridor using pages 44 to 61 of the CBI report — the Cronje phone transcripts, the amounts in rand, the dates. Two rival reporters filed the same story with the wrong number of players named. Mine held, because I had a page number. Here there is no page number. "First association" is a superlative with no definition attached — first in what sense, on which technology, at which level? Until that definition arrives, the claim is marketing, not fact. A clock in my head beats the official timestamp, and right now it says the sentence will take a week to correct and the pilot will take twelve months to judge.

This is also where I should give myself the chance to be wrong, because the biggest error of my forty years has been mistaking silence for evidence.

First: perhaps withholding numbers is the right call. A pilot means not-yet-known. Had MCA declared "we expect a 12 per cent improvement," that number would not have been true — it would have become a benchmark, and every subsequent report would have started from it. My own 2026 win carried that same risk; being early once can be luck, being early repeatedly is a source network. By the same logic, forcing MCA to publish figures prematurely would be wrong.

Second: my indoor-versus-outdoor objection may be overdone. The pilot's scope is explicitly indoor nets, and net coaching is valuable in itself regardless of how much of it transfers to match conditions. If a 19-year-old fast bowler's release point stabilises by two centimetres over three months, that win is real even if it happened indoors. What I object to is not the technology but the expanded claim made on its behalf.

Third: the pilot may not be for players at all, but for coaches — training a coach's eye with data. If so, my "this does not fix conversion" argument softens too, because coaching capacity is an institutional asset that stays after a player leaves. And the most honest point of all: sometimes the ledger supports the consensus. The consensus here is that the pilot is a good thing. I agree. My disagreement is only with the marketing language — turning a bounded, clean, feedback-gated experiment into history by calling it "first."

So note the dates. The formal start of the pilot. The first public feedback — how many coaches, how many players, what changed. The expansion decision — whether this reaches the ground level beyond one academy or remains a press release. My prediction is plain: without a journalist asking, no metric will be published in the next twelve months, and the word "first" will stay on the website untouched.

If Mumbai's selection is ever explained by AI in the next Ranji season, the first question will be: show us what you have. Because selection decisions are public and algorithmic decisions are not. You learn to count the silence before the vote; here the silence is on AI's side, and it will never make the record.

AI will learn in the nets and a human will make the decision — that is fine. The only question is who owns the blame when the decision is wrong: the coach, or the algorithm with no name, no face, and now no number either?