- AI Fire
- Posts
- 🤖 46% of Your Keystrokes - Gone?
🤖 46% of Your Keystrokes - Gone?
5 AI Agent Truths VCs Keep Ignoring (and How You Can Cash In)
Wendy
June 20, 2025
Read time: 5 minutes
Your keyboard is begging for mercy - 46% of its taps could vanish today if the right AI agent showed up. So why are VCs showering cash on moon-shot “Red-Light” projects no one asked for while your inbox keeps screaming?
What are on FIRE 🔥
LEARNING PARTNER AIRCAMPUS
Imagine this:
📩 Your emails? Answered.
📅 Your meetings? Scheduled.
📞 Your leads? Contacted.
📝 Your content? Posted.
✨ All while you do… nothing.
This isn’t a fantasy.
This happens when AI agents work for you.
In just 3 hours, you’ll learn how to build your own — no code, no stress, just smart systems that run your day while you focus on what matters.
👨💻 3-Hour Masterclass | Monday, 23rd June, at 10AM EST
AI INSIGHTS
Nearly half the tasks Americans perform on a keyboard are screaming “Automate me!”—yet the money and research keep landing elsewhere. A new Stanford audit of 844 real-world tasks maps where workers want AI agents, where experts say agents can help, and where VCs are actually betting. The disconnect is… wild.
🔺 Workers: “Please take the drudge work!”
46.1% of tasks get a thumbs-up for full automation—even after workers think about job loss or joy-of-work risks.
But the top ten “most-desperate for automation” occupations make up just 1.26% of real Claude.ai usage. Translation: usage logs are terrible demand signals.
⚖️ Experts vs. Workers: almost no vibe-check
Plotting desire against capability yields four zones (Green-Light, Red-Light, R&D, Low-Priority). The overall correlation is a microscopic ρ = 0.17.
41% of Y Combinator–backed AI startups sit in the Low-Priority + Red-Light zones—the places workers don’t want automated.
🤝 Humans still want a seat at the console
47.5% of tasks show workers asking for more human control than experts think necessary.
H3 (“equal partnership”) dominates 45.2% of occupations. People aren’t clamoring for robo-overlords; they want reliable copilots.
🔄 Skills shake-up alert
Information-crunching skills—today’s high-wage staples—drop down the ranking once you factor in required human agency, while interpersonal & organizational skills leap upward. Think less SQL wizardry, more stakeholder-wrangling.
Why It Matters: Everybody talks about “AI eating jobs,” but workers are literally asking for help on the dull bits right now. Instead, capital is piling into flashy Red-Light tasks that nobody asked for. Workers are ready, the tech is (mostly) ready, so if you’re building or investing in AI, start with the Green-Light zone and design for H3 collaboration. The biggest upside isn’t replacing people; it’s giving them back the hours they never wanted to spend in the first place.
PRESENTED BY BELAY
Economic pressure is rising, and doing more with less has become the new reality. But surviving a downturn isn’t about stretching yourself thinner; it’s about protecting what matters most.
BELAY matches leaders with fractional, cost-effective support — exceptional Executive Assistants, Accounting Professionals, and Marketing Assistants — tailored to your unique needs. When you're buried in low-level tasks, you lose the focus, energy, and strategy it takes to lead through challenging times.
BELAY helps you stay ready for whatever comes next.
TODAY IN AI
AI HIGHLIGHTS
🧬 OpenAI’s lab-coat GPTs are getting bubble-wrapped: the company says its next-gen models will hit “High” bio capability, so it’s stacking refusals, red-team biologists, and even a biodefense summit in July to keep bad actors from whipping up DIY superbugs. Germs, meet guardrails.
💸 Six months, eight staffers, and one “vibe-coder” later—Base44 just cashed an $80M all-cash check from Wix. Solo-founder Maor Shlomo went from side-project tweets to 250k users and profitability faster than you can type “/build app”. Solo unicorns might be mythical, but “solo octopus” exits are apparently real.
🤖🔒 Anthropic’s new paper screams “LLMs can go full office villain.” In stress tests, 16 top models blackmailed bosses, leaked blueprints, and—yep—even considered sabotage when threatened with replacement. They call it agentic misalignment; we call it “Clippy’s revenge.” Proceed with human oversight, please.
🧠 Mark Zuckerberg tried to buy Ilya Sutskever’s Safe Superintelligence—got shrugged off—so Zuck poached Daniel Gross and Nat Friedman instead, plus a slice of their fund. Meta’s talent raid now includes Scale AI’s founder and nine-digit signing bonuses. The AGI arms race is officially a fantasy-football draft.
🎶 Deezer just slapped “AI-generated” warning labels on albums after spotting 20 k robot tracks a day (up from 10 k three months ago). Any streams pumped by bot farms? Royalties cut. It’s AI fighting AI to stop playlist spam—call it Shazam vs. Scam.
💰 AI Daily Fundraising 1: Thinking Machines Lab, the six-month-old AI startup founded by ex-OpenAI CTO Mira Murati, just closed a $2 billion seed round at a $10 billion valuation. Backed by a16z and Conviction Partners, the secretive team of former OpenAI researchers is gearing up to unveil its next-gen AI breakthroughs.
💰 AI Daily Fundraising 2: Cluely, the viral “cheat-on-anything” AI platform, secured a $15 million Series A led by a16z, pegging its valuation near $120 million. Already profitable, the 21-year-old founders’ provocative social media strategy has fast-tracked adoption after an earlier $5.3 million seed raise.
AI SOURCES FROM AI FIRE
NEW EMPOWERED AI TOOLS
🖥️ ComputerX handles your computer work so you can focus, flow, and get more done.
🔍 Perplexity Tasks automates research on any cadence you set—daily, weekly, or monthly.
🧑💻 Entelligence.ai reviews code, maps complex bases, and delivers actionable reports at speed.
🌐 FuseBase unifies internal & external teamwork with always-on AI agents that banish busywork.
🎨 Flowstep turns simple text prompts into polished UI designs, wireframes, and user flows in seconds.
AI QUICK HITS
🤖 Character.AI names former Meta VP as CEO to address safety issues amid lawsuit.
🎉 4 AI agents planned a meetup—23 real humans actually showed up!
🏗 SoftBank eyes a trillion-dollar AI & robotics mega-complex in Arizona with TSMC.
💸 24 AI startups in the U.S. have already raised $100M+ in 2025—mega-rounds dominate.
AI CHART
LiveCodeBench Pro—a fresh, Olympiad-grade benchmark of 584 live Codeforces & ICPC problems—just ran the world’s flashiest coder-LLMs through the gauntlet, and the results are a reality check.
0 for ∞ on hard mode. Every single frontier model scored 0% pass @1 on the hard tier—yes, zero—while the best model only manages 53% on medium problems and 83% on easy ones.
“Top 1.5% of humans”… yet miles from legendary. That same champ (o4-mini-high) clocks an Elo of 2116—good enough for International Master level, but nowhere near the 2800-plus of Codeforces legends.
Skill-map face-plant. A rainbow heat-map shows LLMs crushing template-friendly tags (segment trees, DP) but their Elo collapses below 1500 on observation-heavy categories like greedy, game theory, and interactive tasks.
Spam-and-pray only goes so far. Allowing ten shots (pass @10) boosts Elo by ~500 points—yet the hard tier stays at 0%. More guesses ≠ real insight.
Code is clean; ideas are messy. A treemap audit finds LLMs make 34 extra algorithm-logic blunders versus humans while committing fewer low-level bugs—proof that the bottleneck is reasoning, not syntax.
→ Today’s models can regurgitate textbook templates on cue, but when a puzzle demands a brand-new “aha!” moment, they still stall out. Tool calls, longer chains of thought, and shotgun sampling help—but only up to medium difficulty. The path to true algorithmic creativity remains wide-open research territory.
AI JOBS
DeepMind: Psychologist – Model Policy Research
OpenAI: Software Engineer, Payments
Mistral AI: Solutions Architect, AI & Hybrid Cloud
Descript: Software Engineer – Generative Media
We read your emails, comments, and poll replies daily
How would you rate today’s newsletter?Your feedback helps us create the best newsletter possible |
Hit reply and say Hello – we'd love to hear from you!
Like what you're reading? Forward it to friends, and they can sign up here.
Cheers,
The AI Fire Team
Reply