Scores by kind of claim
Each kind of claim is scored on its own, out of 100: the share that held up, with the range it likely falls in. Fewer resolved claims give a wider range.
-
Factual claims
Statements about the present or the past, checked against the record: did they hold up?
83 77 88 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
Insufficient record: 4 predictions resolved, 5 needed for a score.
-
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
77 65 88
All types combined: 82 (likely 77 to 86, high confidence), from 94 claims: 77 factual claims, 17 promises. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Satya Nadella's claims
5 claims
“I was able to use a model that is using 500B, I was able to use a 5B, and have it really deliver the same outcome, why would I not use that?”
Satya Nadella, Microsoft CEO 4 June 2026 An Interview with Microsoft CEO Satya Nadella About Finding Core Competencies (Stratechery)
In a Land O'Lakes agent example shown at Build, a 5B-parameter model delivered the same outcome as a 500B-parameter model.
Mixed Partly true and partly false in roughly equal measure.
The second judge would not resolve this claim; the verdict shown is the first judge's.
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
Judged as of 2026-06-04. True: Microsoft Build 2026 materials or coverage show a Land O'Lakes agent example in which a model of about 5B parameters delivered the same, or equivalent, outcome as a model of about 500B parameters. Mostly true: the example exists, but the parameter counts differ somewhat (e.g., a small model of a few billion parameters against a frontier model), or results were 'comparable' rather than identical. Mixed: a Land O'Lakes small-model example was shown, but the comparison with the large model is not documented. Mostly false: the example was hypothetical, or involved a different customer or setup. False: no such example was shown, or it showed the small model performing materially worse.
Evidence
- Here's Everything Announced at Microsoft Build 2026 eWeek (The Neuron) · 3 June 2026
“The demo then used Land O’Lakes as the example. The presenter showed how an environment can include skills, knowledge, and tools.”
- Here's Everything Announced at Microsoft Build 2026 eWeek (The Neuron) · 3 June 2026
“The presenter said almost 80% accuracy is not good enough for those tasks, so Microsoft is extending the definition of skills to include rubrics on “what good looks like.””
- Frontier Tuning: Microsoft Build 2026 daily.dev (Microsoft Developer source) · 3 June 2026 Own record
“A demo with Land O'Lakes showed the tuned MAI model achieving over 90% accuracy on complex reporting tasks, estimated at 10x efficiency over baseline models.”
- What’s 🔥 in Enterprise IT/VC #501 What's Hot in Enterprise IT/VC (newsletter) · 6 June 2026
“Microsoft claims that Land-O-Lakes has a 10x more cost efficient model than Open AI’s GPT 5.5.”
Reasoning
Coverage confirms a Land O'Lakes agent example was shown at Build 2026 (E1, E2). A Microsoft Developer summary says a tuned MAI model reached over 90% accuracy on the reporting task at an estimated 10x efficiency over baseline models (E3). A newsletter reports Microsoft's claim that the model was 10x more cost efficient than GPT 5.5 (E4). No source gives about 5B or 500B parameter counts or shows the two models reaching the same outcome, so the specific comparison in the claim is not documented.
Cite this claim
Satya Nadella, “In a Land O'Lakes agent example shown at Build, a 5B-parameter model delivered the same outcome as a 500B-parameter model.” (said 4 June 2026). Verdict: mixed. Reliability API, claim c_11bd759401ec, method 0.1.0+28da144b. https://epistacklabs.com/claims/c_11bd759401ec
Claims without a verdict
Is a verdict here wrong, or do you speak for Satya Nadella? Reply on the claim's own page, with the form at its foot (how replies work). Verdicts are made automatically; how verdicts and scores are made.