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
43 claims
“You now have a full local agentic loop, you can give it tools access and build fully onboard agentic applications without having to round trip to the cloud.”
Satya Nadella, Microsoft CEO, keynote 2 June 2026 Satya Nadella Microsoft Build 2026 keynote transcript (Microsoft Source)
Developers can now use a full local agentic loop with tool access to build on-device agentic applications without round trips to the cloud.
Mostly true True in substance, wrong in a detail.
The second judge would not resolve this claim; the verdict shown is the first judge's.
The evidence is the person's or their organisation's own records.
Said for an organisation they lead or work for; it counts on both records.
What would count as true or false
Judge as of 2 June 2026. True: Microsoft had released a capability that developers could use by that date, for example in Foundry Local, Windows AI or a similar on-device runtime, that runs a full agentic loop locally with tool or function calling, letting agentic applications run entirely on the device without cloud calls. Mostly true: the capability is available, but only in preview, or with minor limits such as specific hardware or a subset of tools. Mostly false: the capability was only announced as coming later, or the core loop still requires the cloud. False: no such local agentic capability existed.
Evidence
- Build 2026: Furthering Windows as the trusted platform for development Microsoft Windows Developer Blog (blogs.windows.com) · 2 June 2026 Own record
“A new generation of on-device models – Aion 1.0 Instruct and Aion 1.0 Plan in preview”
- Build 2026: Furthering Windows as the trusted platform for development Microsoft Windows Developer Blog (blogs.windows.com) · 2 June 2026 Own record
“Introducing new on-device SLMs – Aion 1.0 Instruct, a smaller, faster and smarter on-device SLM, and Aion 1.0 Plan, a reasoning and tool-calling model that enables fully local agentic capabilities, available in the coming months.”
- What's new in Microsoft Foundry | April 2026 | Microsoft Foundry Blog Microsoft Foundry Blog (devblogs.microsoft.com) · 12 May 2026 Own record
“Foundry Local (generally available, GA): Local model inference is production-ready on Windows, macOS on Apple Silicon, and Linux x64.”
- Use tool calling with Foundry Local - Foundry Local Microsoft Learn · 3 November 2026 Own record
“Foundry Local can make use of tool calling, a technique where you prompt the model with definitions of available tools that together with a text prompt, allow the model to work out which tools should be called and with what input data.”
- Microsoft Build 2026: Everything Microsoft Announced The Neuron · 2 June 2026
“Together, Satya described them as enabling a full local agentic loop, where developers can give agents tool access and build agentic applications without a round trip to the cloud.”
- Build 2026: Furthering Windows as the trusted platform for development Microsoft Windows Developer Blog (blogs.windows.com) · 2 June 2026 Own record
“Aion 1.0 Plan, a reasoning and tool-calling model that enables fully local agentic capabilities, available in the coming months.”
- Build 2026: Furthering Windows as the trusted platform for development Microsoft Windows Developer Blog (blogs.windows.com) · 2 June 2026 Own record
“A new generation of on-device models – Aion 1.0 Instruct and Aion 1.0 Plan in preview”
- Use tool calling with Foundry Local - Foundry Local Microsoft Learn · 3 November 2026 Own record
“Foundry Local can make use of tool calling, a technique where you prompt the model with definitions of available tools that together with a text prompt, allow the model to work out which tools should be called and with what input data.”
- Build Personal AI Agents on Windows PCs with New Tools from Microsoft and NVIDIA | NVIDIA Technical Blog NVIDIA · 2 June 2026
“At Microsoft Build, Microsoft announced a set of security primitives to allow agents to execute code, operate on files, and orchestrate tasks across systems with built-in identity and policy execution.”
- What's new in Microsoft Foundry | May 2026 | Microsoft Foundry Blog Microsoft Foundry Blog (Microsoft) · 31 May 2026 Own record
“Three related Foundry Labs from Microsoft Research released this month: MagenticLite, the local agentic app;”
- Advanced Function Calling and Multi-Agent Systems with Small Language Models in Foundry Local | Microsoft Community Hub Microsoft Community Hub · 15 January 2026 Own record
“Currently, this native function calling capability is available for the Qwen 2.5 family of models in Foundry Local.”
- Use tool calling with Foundry Local - Foundry Local Microsoft Learn · 3 November 2026 Own record
“Foundry Local can make use of tool calling, a technique where you prompt the model with definitions of available tools that together with a text prompt, allow the model to work out which tools should be called and with what input data.”
Reasoning
The Aion 1.0 Plan model that Nadella linked to the "full local agentic loop" is not clearly settled as available on 2 June 2026. The same Microsoft post labels it "in preview" (E7) but also says it will be "available in the coming months" (E2, E6). Separately, Microsoft's own records show that a local tool-calling capability existed before that date. Foundry Local had native function calling for local models by January 2026, but only for the Qwen 2.5 model family (E11). It was declared generally available for production local inference in April 2026 (E3), and MagenticLite, a local agentic app, was released in May 2026 (E10). Since a usable local tool-calling path existed but was limited to certain models, and the headline model was preview or still to come, the claim is mostly true; the support comes from Microsoft's official records, and the independent source (E5) only reports what Nadella said.
Cite this claim
Satya Nadella, “Developers can now use a full local agentic loop with tool access to build on-device agentic applications without round trips to the cloud.” (said 2 June 2026). Verdict: mostly true. Reliability API, claim c_c16c35bb8653, method 0.1.0+28da144b. https://epistacklabs.com/claims/c_c16c35bb8653
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.