Microsoft put SharePoint tools in Work IQ into public preview on 3 November, and says that in its internal evaluations they raised Work IQ’s task pass rate from 70.1% to 84.6% while using 16% fewer tokens. By our arithmetic, that is 14.5 percentage points, and about 48% fewer failed tasks.
Jeff Teper, whose X profile describes him as EVP of Apps & Agents at Microsoft, posted on X that “Microsoft Copilot and agents delivers better outcomes \ $ vs. ChatGPT and Claude”. The Microsoft blog post he linked does not compare Copilot with ChatGPT or Claude.
Business Pill · Usage-Based Pricing
An evergreen Business Engineer explainer on usage-based pricing. It teaches the general idea only and does not discuss Microsoft or SharePoint.
The key insight:Microsoft says its SharePoint tools raised Work IQ’s internal task pass rate from 70.1% to 84.6% with 16% fewer tokens, 14.5 percentage points and about 48% fewer failed tasks by our arithmetic; the post compares Microsoft’s own setups, not Copilot against ChatGPT or Claude.
According to the SharePoint team’s post by Patrick Rodgers, the preview “adds more than 60 native SharePoint operations through the Work IQ MCP server”, so “Copilots and agents can find, understand, create, organize and share SharePoint content”. Microsoft calls Work IQ “the intelligence layer behind Microsoft 365 Copilot and agents”.
The post says the tools are “the same tools” used in Copilot in SharePoint, and that they are now available to “ANY agent: GitHub Copilot CLI, VS Code, or the agent you’re building yourself”. They are also rolling out to Microsoft Copilot Studio.
Work IQ task pass rate in Microsoft’s internal evaluations, without and with the new SharePoint tools: 70.1% and 84.6%, with 16% fewer tokens. Data: Microsoft SharePoint blog, 3 November 2026. The 14.5-point gain is our arithmetic.
What Was Posted on X
Microsoft Copilot and agents delivers better outcomes \ $ vs. ChatGPT and Claude. Why 1. We optimize model usage. 2. We deliver AI native APIs and data enrichment.
Today’s latest news is we are bringing the #SharePoint AI tools into Work IQ for a range of solutions.
Jeff Teper, whose X profile describes him as EVP of Apps & Agents at Microsoft, claims Copilot and agents deliver better outcomes per dollar than ChatGPT and Claude, citing optimized model usage and AI-native APIs, and links the SharePoint tools announcement.
The Evidence Microsoft Gives
The post says: “In our internal evals, adding SharePoint tools raised Work IQ’s task pass rate from 70.1% to 84.6% while using 16% fewer tokens in our testing”. It adds: “When compared against directly accessing Graph, we see a lift in success rate up to 77% depending on task”.
Both comparisons are within Microsoft’s own stack: Work IQ with and without the SharePoint tools, and the tools against direct Graph access. Teper’s post gives two reasons for his outcomes-per-dollar claim: “We optimize model usage” and “We deliver AI native APIs and data enrichment”.
How It Is Governed
Microsoft says every call “runs as the signed-in user”, and that two new scopes support least-privilege access: WorkIQ.Read for searching, reading and summarising, and WorkIQ.Act for creating, updating or moving content. It says SharePoint tools are “the first Work IQ provider to support these permissions”.
As we read the post, Microsoft’s argument is that the expertise should sit in the tool rather than in the prompt: purpose-built operations mean the model spends fewer tokens working out how to act.
As we read Teper’s claim, outcomes per dollar is the metric Microsoft wants buyers to use; the published evidence covers outcomes and tokens inside Microsoft’s stack, and prices against rivals are not part of it.
Jeff Teper, on X
“Microsoft Copilot and agents delivers better outcomes \ $ vs. ChatGPT and Claude.”
Three Implications
FOR IT ADMINSMicrosoft says admins enable Work IQ for the tenant and choose which SharePoint paths and actions are allowed through tenant policy.
FOR AGENT BUILDERSMicrosoft says the tools are available to any agent through the Work IQ MCP server, including GitHub Copilot CLI and VS Code.
FOR BUYERS COMPARING ASSISTANTS Teper’s outcomes-per-dollar comparison with ChatGPT and Claude comes without published data in his post or the blog.
The framework says of the token: “It is simultaneously the unit of cognition the model produces, the unit of compute the data center serves, the unit of price the lab charges, and the unit of value the enterprise extracts”.
As we read it, Microsoft is arguing on two of those units at once: fewer tokens per task and more tasks completed. The framework predates this launch and does not discuss Microsoft.
The pass-rate and token figures are Microsoft’s internal evaluations; the post does not give the number of tasks or the models used. Teper’s outcomes-per-dollar comparison with ChatGPT and Claude is his claim, and neither his post nor the blog post publishes data for it.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Microsoft says its new SharePoint tools for agents raised Work IQ’s internal task pass rate from 70.1% to 84.6% with 16% fewer tokens, about 48% fewer failures by our arithmetic. Jeff Teper used the launch to claim better outcomes per dollar than ChatGPT and Claude; the evidence Microsoft published compares its own tools with and without the new layer.
A note on sourcing. The preview, the pass-rate and token figures and the governance details are from Microsoft’s SharePoint blog post of 3 November 2026, based on Microsoft’s internal evaluations. The outcomes-per-dollar claim is Jeff Teper’s, from his post on X. The 14.5-point gain and the share of failures removed are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
