Microsoft Cancelled Its Own AI Tool. That Should Tell You Everything.

Microsoft Cancelled Its Own AI Tool. That Should Tell You Everything.

Reflect w/ Ed Fassio — where AI tells the stories that matter. Fortune put Microsoft on its June cover with a headline worth sitting with: "Microsoft lost its way in the AI race. Can Copilot get it back on course?" Internally, Microsoft cancelled Claude Code licenses across 5,000 of its own engineers — per-engineer costs hit $500 to $2,000 a month with no measurable productivity return. Hale and Julius dig into what happened and why it matters for every enterprise running a Copilot deployment right now. The lesson isn't about Claude Code specifically. It's about what happens when adoption without governance meets fiscal year-end: 79% of organizations report AI adoption challenges, 46% say initiatives haven't met expectations, and 95% of GenAI pilots fail to scale. The companies making AI work financially share one trait: they defined outcomes before rollout, not after the bill arrived. Your Move: Build the measurement framework before expanding the budget. Set cost ceilings. Review the math at 90 days, not 12 months. And ask your team one question — if the budget doubled next quarter, could we prove we'd double the return? — Reflect w/ Ed Fassio | reflectpodcast.com Send us Fan Mail [https://www.buzzsprout.com/2414886/fan_mail/new] Support the show [https://www.buzzsprout.com/2414886/support] LISTEN TO MORE EPISODES: https://www.reflectpodcast.com [https://www.reflectpodcast.com]

Sam Altman Said He Was Wrong. Here's Why That Should Worry You More.

Reflect w/ Ed Fassio — where AI tells the stories that matter. Eight days after filing OpenAI's IPO paperwork, Sam Altman publicly walked back his AI jobs apocalypse prediction. Dario Amodei did the same thing the same week. Meanwhile Uber burned its entire $3.4 billion AI budget in four months with nothing measurable to show for it. And NVIDIA's own VP said AI compute now costs more than the employees it was supposed to replace. Julius and Hale break down what's actually in the receipts: the Uber COO who can't draw a line between AI spending and output, the Microsoft cancellation that reveals the real cost of uncapped usage, and why the Altman reversal isn't the all-clear signal it looks like. The companies winning with AI right now aren't the ones who believed the hype the hardest. They're the ones who built governance tight enough to survive the reckoning. Your Move: Don't read the Altman reversal as an all-clear. Build the measurement framework before you expand the budget. And watch the product roadmap — not the narrative. — Reflect w/ Ed Fassio | reflectpodcast.com Send us Fan Mail [https://www.buzzsprout.com/2414886/fan_mail/new] Support the show [https://www.buzzsprout.com/2414886/support] LISTEN TO MORE EPISODES: https://www.reflectpodcast.com [https://www.reflectpodcast.com]

AIRD: The New Diagnosis Nobody Wants to Talk About

Reflect w/ Ed Fassio — where AI tells the stories that matter. University of Florida researchers just gave it a name: AI Replacement Dysfunction — AIRD. It's a clinical stress pattern showing up in workers whose jobs are perceived to be at risk from AI. Not general anxiety. A specific, recognized dysfunction: persistent stress, disrupted sleep, difficulty concentrating, reduced engagement. The numbers are hard to sit with. 1 in 4 employees say AI is already hurting their mental health. 72% feel pressure to work through mental health challenges — up 10 points from last year. 51% reported crying at work in the last 30 days. And here's what makes it complicated: AIRD hits hardest in organizations doing adoption the right way. The transparency that's supposed to build trust actually triggers the anxiety in employees who are honest enough with themselves to understand what's coming. The 42-year-old marketing manager watching AI generate a month of content in 45 minutes. The paralegal who spent seven years mastering contract review. Their expertise — the thing their identity was built around — is being commoditized in real time. Your Move: If you're feeling AIRD, separate your identity from your job description. Your judgment, relationships, and domain knowledge aren't being automated — the routine layers are. If you're a manager, check in with your team this week about how they're actually feeling — not about productivity. The data says 1 in 4 of them are struggling. And if you're an organizational leader: running AI training while privately planning layoffs destroys trust in ways that don't recover. AIRD is a business risk. Treat it like one. — Reflect w/ Ed Fassio | reflectpodcast.com Send us Fan Mail [https://www.buzzsprout.com/2414886/fan_mail/new] Support the show [https://www.buzzsprout.com/2414886/support] LISTEN TO MORE EPISODES: https://www.reflectpodcast.com [https://www.reflectpodcast.com]

Stalled at Week Eight: Why Most Enterprise AI Deployments Never Actually Land

Reflect w/ Ed Fassio — where AI tells the stories that matter. Tuesday we talked about Mustafa Suleiman's 12-to-18 month window. Today Julius and Hale get into why most organizations are going to miss it — not because the tools don't work, but because of what happens between week six and week twelve of every enterprise AI rollout. The pattern is consistent whether you're deploying Microsoft Copilot, Salesforce Einstein, or a custom agentic workflow. Organizations launch with energy. Kickoffs happen. Memos go out. And then momentum quietly dies. Adoption plateaus at 15-20% of intended users. Everyone else drifts back to how they were working before. The root cause isn't technology. It's governance treated as a setup task instead of an ongoing process. And the fix is specific: a dedicated governance owner (not a committee), feedback loops from day one, strategic use case selection, and manager enablement — not just end-user training. Your Move: Pull your real usage data right now. Not license count — active users completing a workflow task at least twice a week. If you're under 30%, you have an adoption problem governance alone won't fix. Go back to use case basics. If you're over 50% but hitting friction, that's a solvable infrastructure problem. Either way, the playbook exists. The failure modes are documented. There's no excuse to walk into a deployment in 2026 without a governance framework from day one. — Reflect w/ Ed Fassio | reflectpodcast.com Send us Fan Mail [https://www.buzzsprout.com/2414886/fan_mail/new] Support the show [https://www.buzzsprout.com/2414886/support] LISTEN TO MORE EPISODES: https://www.reflectpodcast.com [https://www.reflectpodcast.com]

The $20 Subscription vs. the $120K Salary: What Microsoft's AI Chief Just Said Out Loud

Reflect w/ Ed Fassio — where AI tells the stories that matter. Microsoft's Chief AI Officer Mustafa Suleiman just said it out loud: most white-collar jobs — lawyers, accountants, marketers — could be fully automated within 12 to 18 months. Not a decade from now. By the end of next year. Julius and Hale break down what that prediction actually means when you look at the numbers. A fully-loaded mid-career accountant costs $110K–$130K a year. A Copilot license costs $360. That's not a philosophical debate about the future of work — it's a fiduciary conversation happening in boardrooms right now. But Tuesday on Reflect is about what smart organizations are actually doing. And the answer isn't panic. It's reframe. The organizations winning right now aren't asking 'will AI replace my team?' They're asking 'which tasks are automatable, and what do I want my people doing instead?' BCG says 50–55% of US jobs will be reshaped — not eliminated — in the next two to three years. The people who move now build the competency before it becomes a crisis. Your Move: Run a task audit — not a job audit. List every recurring task your team does weekly and ask honestly which ones AI could handle at 80% accuracy. Then identify your highest-leverage people and start moving them toward the AI-augmented version of their role. Don't wait for a top-down mandate. The window is 12 to 18 months. Start your clock. — Reflect w/ Ed Fassio | reflectpodcast.com Send us Fan Mail [https://www.buzzsprout.com/2414886/fan_mail/new] Support the show [https://www.buzzsprout.com/2414886/support] LISTEN TO MORE EPISODES: https://www.reflectpodcast.com [https://www.reflectpodcast.com]