# Evan Shapiro

Evan Shapiro writes about helping businesses use AI with shared, reliable company data and definitions. Their posts focus on making AI useful for business decisions, rather than leaving it as a tool used by a few individuals.

- Profile: https://b2binfluencer.marketing/creators/evan-shapiro/
- LinkedIn: https://www.linkedin.com/in/evan-shapiro-b2517ab3
- LinkedIn followers: 3,717
- Topics: artificial-intelligence, ai-adoption
- Based on 9 dated posts, latest 2026-10-08

## Summary

Most useful for explaining why AI seats and licences do not produce trusted company-wide use.

- Best for: B2B SaaS AI assistant or analytics brand considering a paid talk or advisory session.
- Not for: Consumer app or retail brand needing paid social video with guaranteed mass reach.
- Check first: Confirm recent audience data, fees, availability, and any competing interest with MIRA.

### Takeaways

- One power user’s AI is not a company capability.
- Paid AI seats go unused because answers can’t be traced, not because access is missing.
- Fix the decision problem before buying infrastructure.

## Main ideas

- **One power user’s AI is not a company capability**: Evan Shapiro argues that general AI is a tool an individual uses, while organisational AI is a capability a company owns. The difference becomes clear as soon as more than one person relies on it. In the SaaS businesses they describe, a capable user in revenue operations or sales leadership connects AI to the CRM and automates their own reporting, while the rest of the team still reads dashboards and waits for answers. From the outside, the company looks like an early adopter, but the capability lives in one person’s account. The problem compounds as use spreads: different people’s AI computes the same quantity using different definitions, and nobody notices until the numbers disagree in front of a board. Their proposed fix is to let anyone ask a question and draw answers from agreed sources and shared definitions, so two people asking the same thing get the same number.
- **Paid AI seats go unused because answers can’t be traced, not because access is missing**: Shapiro separates distribution from adoption. Copilot already sits inside Word, Excel, Outlook and Teams, with nothing to install, yet the post reports that fewer than 4.5% of those roughly 450 million seats pay for it, and that only 20 to 30% of paying users reportedly use it in a given week. Their explanation is a verifiability gap: AI confidently answers a question about the business, someone asks where the number came from, no source can be shown, and the tool quietly becomes something used for drafting emails. In this account, general AI queries data without understanding the business. Software that draws on the CRM, call recordings and management accounts — and can show where an answer came from — survives a board meeting.
- **Fix the decision problem before buying infrastructure**: Reporting from Reading Football Club’s AI summit, Shapiro argues that organisations, football clubs included, are not short of data: it already sits in the finance system, the CRM, ticketing and the spreadsheet someone updates every Monday morning. The hard part is identifying where time is being lost, which decisions are difficult to make, and where an answer exists but takes a day of pulling reports to reach. Once that is clear, the infrastructure conversation becomes focused because you know what you are building for. The panel Shapiro singles out, "How the biggest companies are investing in football", featured speakers from NVIDIA and HUMAIN and centred on the infrastructure needed to run AI at scale — a view presented alongside their own, not as their position.

## Work to read

- [General AI is a tool an individual uses; organisational AI is a capability a company owns](https://www.linkedin.com/posts/evan-shapiro-b2517ab3_generalai-organisationalai-saas-activity-7495796760112697344-V85S)
- [Microsoft put Copilot in front of 450 million seats and fewer than 4.5% pay for it](https://www.linkedin.com/posts/evan-shapiro-b2517ab3_microsoft-put-copilot-in-front-of-450-million-activity-7493565778827366401-GqWU)
- [On Friday I was at Reading Football Club’s AI summit](https://www.linkedin.com/posts/evan-shapiro-b2517ab3_on-friday-i-was-at-reading-football-clubs-activity-7510598048121573376-o2Ep)

## Brand relationships

- **MIRA** (disclosed_partnership)

## Partnerships

The three posts contain no sponsorship offer, rate card, media kit or published partnership terms. The only commercial activity described is Evan Shapiro’s own business, MIRA, which they present as helping business leaders see the reality of their performance through data-driven decisions rather than guesswork. When the posts invite a response, it’s tied to that business: readers are asked how AI capability lives inside their company or what a club is actually trying to decide. They don’t invite paid placements, sponsored posts or brand collaborations. A buyer should ask Shapiro directly how to open a partnership conversation. Partnerships would also touch on the same territory the writing covers: AI adoption, shared definitions and trusted answers inside a company, and the operational decisions that make AI infrastructure worth funding.

## Alternatives

- James Wilkinson: A better choice when the client sells to clubs, hospitality or small service operators and the session needs to explain everyday software adoption rather than enterprise AI strategy. One post explains why a CRM rollout was hard: the blocker was the team's fear of a new way of working and the learning curve, not the software. It also describes the training material built to help people through it. The other shows AI cutting the cost of bespoke website work, with build time down to weeks, sites search-optimised from the start and connected to a CRM, and posts written in the client's own tone. For a small-team audience, that's closer to their reality than a talk about company-wide AI capability. The change-management detail comes from doing the work, not from a framework.
- Didi Gurfinkel: A better choice when the buyer is the office of the CFO and the AI story turns on financial data being consolidated, mapped and auditable. The writing makes the same trust point as Evan Shapiro—that an answer is only usable if it is grounded in real numbers—but grounds it in specifics: several ERPs in one group, some local books kept by outside firms, and the difference between an AI tool stitching data together when the mappings are wrong and an output that is reliable and audit-ready. One post describes a finance-native ticketing system where each request carries its budget, commitments and forecast impact, with permissions keeping AP, AR, Payroll and FP&A data apart, and quotes the finance lead at the company involved. That gives a client a named, checkable example close to their own data problem.
- Lauren B. Jones: A better choice when the session is a hands-on enablement workshop for a functional team—recruiters or back-office staff—rather than a positioning talk for leadership. The material is close to a ready run-of-show: five named Claude uses, each a concrete move, from giving every client account its own project so past job specs and rate cards stay loaded, to turning one intake call into five deliverables, bottling a top performer's writing style into a reusable skill, connecting tools so a brief is waiting on Monday morning, and keeping a human on the hiring decision. The same post advertises two free live sessions on Copilot and Claude, direct evidence of training being delivered to this audience.

---
AI-assisted summary of public posts. Audience demographics are not available. Engagement appears only where we checked recent posts. Method: https://b2binfluencer.marketing/methodology/