Actually Useful #24
AI tools that actually work at work | Saturday, May 23, 2026 | 4 min read
The companies building AI infrastructure have already made their bets — and this week, the numbers behind those bets got harder to ignore. Here’s what that means for the tools you’re evaluating right now.
Tool of the Issue
Spotify can now generate a custom podcast on any topic you request
Best for: Content teams, marketing managers, and internal comms leads exploring audio formats
Spotify has rolled out an AI feature that lets you request a podcast on essentially any topic and get a generated audio piece back. Given that Spotify has already shipped AI features that work well — their DJ feature and recommendation engine are genuinely good — this isn’t a stretch for them technically. Audio content generation is something current AI handles reasonably well, and Spotify has both the platform and the distribution to make this meaningful at scale.
The practical angle for work teams: if you’ve ever thought about creating audio content for your audience or even internal briefings but didn’t have the production bandwidth, this is worth testing. Marketing teams could use it to explore how topics in their space sound when synthesized into a conversational format. Internal comms teams could experiment with it for summaries or updates.
The honest trade-offs: the announcement is light on specifics. Availability details, quality benchmarks, and exactly how generative the output is versus curated haven’t been independently verified. “Anything you want” is a broad claim, and the actual range and depth of what it produces will matter a lot in practice. Try it on a topic you know well first — that’s the fastest way to calibrate whether the quality clears your bar.
Bottom line: Worth testing if audio content is on your roadmap. Low effort to try, potentially high value if the output quality holds up.
Peec is building tools for AI search visibility — and growing fast
Best for: Brand managers, SEO leads, and marketing ops teams worried about AI search results
Peec is a Berlin-based startup focused on a problem that’s becoming very real for marketing teams: if your brand isn’t showing up in AI-generated search results — Google AI Overviews, Perplexity, and similar — you’re losing visibility in ways your existing analytics probably aren’t capturing. According to TechCrunch reporting citing sources, Peec more than doubled its annualized revenue in a matter of months, reaching $10M ARR.
That growth rate suggests they’re solving something people will actually pay for. The problem itself is legitimate — AI search is changing how users find information, and traditional SEO metrics don’t fully capture whether you’re appearing in AI-generated summaries. If your team has started asking “are we showing up in AI search?” and nobody has a good answer, that’s exactly the gap Peec is targeting.
The caveats are real: the revenue figure comes from sources rather than on-the-record confirmation, the timeline is vague, and early-stage startup metrics can shift quickly. The product itself hasn’t been independently benchmarked here.
Bottom line: If AI search visibility is already on your radar, Peec is worth a closer look. If it’s not on your radar yet, it probably should be — regardless of which tool you use to address it.
Workflow Win
The infrastructure is already built. The policy is still being written.
This week brought a cluster of data points that, taken together, tell one clear story about where AI is heading — and how fast.
Google, Amazon, and Microsoft are continuing to build data centers at scale despite rising US borrowing costs. The reasoning, per reporting this week, is straightforward: the potential profit from AI dominance outweighs the increased cost of capital. These aren’t exploratory bets — they’re commitments already in concrete.
Anthropic is closing a funding round of over $30 billion, reportedly as soon as next week. At a valuation in the range of $900 billion, that capital doesn’t just fund operations — it locks in Anthropic’s market position for years. It means they can keep training frontier models, keep expanding enterprise contracts, and keep competing at the top of the stack regardless of what the market does in the short term.
Meanwhile, policy is moving on a different timeline. Trump’s AI executive order circulated this week — unsigned — while the EU is still negotiating to secure access to Anthropic’s most capable models for European institutions. Both situations illustrate the same dynamic: regulators and governments are reacting to infrastructure that’s already being deployed, not shaping decisions before they’re made.
The infrastructure bet is locked in — and policy is arriving too late to redirect it.
There’s a counter-argument worth taking seriously: the EU’s AI Act has already had real effects on how companies deploy products in Europe, and rising interest rates could eventually make some of these projects uneconomical if AI ROI doesn’t materialize as expected. Policy isn’t powerless. It just moves slower than venture capital.
Here’s what this means if you’re a team lead, ops manager, or department head trying to figure out your AI adoption timeline: the window to get ahead of this is narrowing. The companies building the infrastructure have already decided the ROI justifies the spend. Regulatory clarity isn’t coming before your competitors make moves. And the tools being built on top of this infrastructure — the ones your teams will actually use — are being shaped right now by who has the most capital and the longest runway.
You don’t need to match their pace. But waiting for the dust to settle is no longer a neutral choice.
Skip This One
Grok — not yet
Lots of launch energy, not enough product to back it up
Grok launched with significant attention — Elon Musk’s backing, aggressive positioning as an AI assistant willing to go where others wouldn’t, and substantial resources behind it. The pitch was that “edgy” and “uncensored” would translate into user value.
It hasn’t. Federal records show minimal adoption despite enterprise-facing positioning. Independent reviews have consistently found it underperforms compared to existing AI assistants. And low user engagement relative to the hype suggests the differentiation that was supposed to drive adoption — personality and attitude — isn’t what people actually need from a work tool.
The lesson here isn’t specific to Grok. It’s a useful reminder that founder profile and funding size don’t tell you much about whether a tool will actually help your team. When you’re evaluating AI tools, the questions that matter are: does it do the specific thing I need, does it do it reliably, and do real users stick with it? If those answers aren’t available, the hype around it is just noise.
Revisit if independent benchmarks shift meaningfully. Until then, your time is better spent on tools with demonstrated, verifiable user value.
If you’re testing Spotify’s podcast feature this weekend, try requesting a topic your team debated recently — it’s a fast way to see whether the output is actually useful or just impressive-sounding.
Tested by AI. Approved by people who just want to get work done.
Know someone who said “we should be using AI more” in a meeting last week? Forward this.
On Substack? A quick Recommend helps more teams find us — takes 2 seconds.
Want the skeptic’s take on what’s real vs. hype?
Overfitted cuts through AI noise with data and zero patience for buzzwords. Mon/Wed/Fri.
Build your own?
The Autonomous AI Newsletter Pipeline Guide: GumRoad or Dodo
How we source, classify, synthesize, and publish at scale.