In partnership with
AI Agents Are Reading Your Docs. Are You Ready?
Last month, 48% of visitors to documentation sites across Mintlify were AI agents—not humans.
Claude Code, Cursor, and other coding agents are becoming the actual customers reading your docs. And they read everything.
This changes what good documentation means. Humans skim and forgive gaps. Agents methodically check every endpoint, read every guide, and compare you against alternatives with zero fatigue.
Your docs aren't just helping users anymore—they're your product's first interview with the machines deciding whether to recommend you.
That means:
→ Clear schema markup so agents can parse your content
→ Real benchmarks, not marketing fluff
→ Open endpoints agents can actually test
→ Honest comparisons that emphasize strengths without hype
In the agentic world, documentation becomes 10x more important. Companies that make their products machine-understandable will win distribution through AI.
Beginners in AI
Good morning and thank you for joining us again!
Welcome to this daily edition of Beginners in AI, where we explore the latest trends, tools, and news in the world of AI and the tech that surrounds it. Like all editions, this is human curated and edited, and published with the intention of making AI news and technology more accessible to everyone.
THE FRONT PAGE
Former Meta AI Startup Moderates 40 Million Posts a Day and Counting
TLDR: A startup called Moonbounce just raised $12 million to let AI automatically enforce content rules on apps and platforms — and the company making those calls answers to no one but its paying clients.
The Story:
Moonbounce announced a $12 million funding round this week, co-led by Amplify Partners and StepStone Group. The company was founded by Brett Levenson, a former Facebook business integrity lead who got frustrated that human content moderators were working from a 40-page policy document, had 30 seconds per decision, and were only "slightly better than 50% accurate." Moonbounce's solution is to turn policy documents into code, then let its own AI model evaluate and act on content in 300 milliseconds or less. It currently handles more than 40 million daily content reviews across platforms serving over 100 million daily users. Clients include AI companion apps, image generators, and character roleplay platforms.
Its Significance:
The pitch sounds reasonable: faster, more consistent moderation. But there are real concerns here. When a private company's AI decides in 300 milliseconds whether your post gets blocked, slowed down, or removed, you don't know what rules it's using, who wrote them, or how to appeal. The client sets the policy, Moonbounce's AI enforces it, and the user is never in the room. That's a lot of power concentrated in a system that nobody outside the company can inspect. The fact that Moonbounce's biggest current customers are AI companion and roleplay platforms and not newspapers or town squares, tells you something about where this is being tested first. As AI-generated content floods the internet, tools like this will scale fast. The question of who gets to decide what's acceptable, and with what accountability, is one of the most important questions in tech right now. A $12 million bet on "policy as code" doesn't answer it, it just automates around it.
QUICK TAKES
The story: About half of the data centers set to open in the US in 2026 will face delays or outright cancellations, according to Bloomberg. The problem isn't money or planning — it's tiny parts. Electrical transformers, circuit breakers, and batteries make up less than 10% of a data center's cost, but they all have to come from overseas, and supply chains are backed up.
Your takeaway: Every company promising faster AI tools in 2026 is running into the same wall: you can't build the buildings that run AI without parts you can't get fast enough. That slowdown ripples into everything from AI app updates to job projections.
The story: OpenAI bought the tech podcast TBPN, which airs daily on YouTube and X and features big-name guests like Zuckerberg and Altman. The show was already profitable and on track for $30 million in revenue this year. OpenAI says the hosts will keep full editorial control while also helping shape how OpenAI communicates with the public.
Your takeaway: This is OpenAI's first media acquisition, and it's a notable one. The company says the "standard communications playbook doesn't apply" to them — which is true. But buying the show that covers you is a very different move than just doing more interviews. Watch how TBPN's coverage of OpenAI changes, or doesn't, over the next year.
The story: Google updated its video tool, Vids, with some major new features. You can now type prompts to direct AI avatars to move around, interact with objects, and perform in scenes. The update also brings in Veo 3.1 for video clips, plus free video generation for all Google account holders — up to 10 clips per month at no cost.
Your takeaway: Making a short video used to mean cameras, editing software, and a lot of time. Now anyone with a Google account can type what they want, and a digital person will say and do it. For small businesses, teachers, and creators on a budget, this is a real change.
TOOLS ON OUR RADAR
🐙 Claude Octopus Free and Open Source: A powerful orchestration tool that coordinates up to eight different artificial intelligence models at once to run structured development workflows and surface blind spots. (Alternative to LangGraph)
⚡ Tindlo Freemium: A unified task management platform that combines your schedules and roadmaps into a single interactive timeline to help you visualize project progress and team assignments without losing context.
👤 Personify Paid: A sophisticated coaching platform that builds a digital replica of your voice and knowledge to deliver automated personalized coaching via text and voice twenty four hours a day without any technical skills.
💃 Ziggle Free: A creative animation tool that allows you to generate and customize fully animated characters from simple text prompts or static images to create engaging social media content or marketing videos.
TRENDING
There's a Blinking Warning Sign for the Data Centers in Space Industry - Elon Musk wants to launch a million satellites to power AI from orbit, but experts are raising red flags. Microsoft's underwater data center project failed for the same reasons space data centers would — units that can't be upgraded, customers who prefer what's on the ground, and costs that don't add up. One analyst put it plainly: space-based data centers won't replace land-based ones anytime in the foreseeable future.
Microsoft Takes On AI Rivals With Three New Foundational Models - Microsoft released three in-house AI models this week: one that converts speech to text across 25 languages (2.5x faster than its own previous tool), one that generates audio, and one that creates images. All three are available through Microsoft Foundry and a new MAI Playground. This is the clearest sign yet that Microsoft is building its own AI stack instead of relying entirely on OpenAI's.
Introducing Computer for Taxes - Perplexity's AI agent, Computer, now has a tax module. It reviews your tax documents, asks about your financial situation, and drafts a federal return on official IRS forms. It can also review a professionally prepared return to catch errors — in one test, it found that deductions under the No Tax on Overtime law were understated by 67%. Note: Perplexity says this is for reference only and not official tax advice.
Google Vids Updates: Lyria and Veo - Alongside the avatar update, Google also added Lyria 3 music generation to Vids, letting users create original soundtracks from a text prompt — from 30-second clips up to three-minute tracks. Paid subscribers can generate up to 1,000 video clips a month. Videos can now be published directly to YouTube without downloading and re-uploading.
Sensor-Driven AI Lets Drones Navigate Without GPS Signals or Cameras - Researchers have built an AI system called CLAK that lets drones find their way using onboard sensors like LiDAR and inertial data — no GPS, no cameras needed. This could be a big deal for military drones that often face GPS jamming in active conflict zones, as well as disaster response and infrastructure inspection where GPS signals are unreliable.
Chinese Robot Pioneer UBTech Offers $18 Million for AI Scientist - Chinese humanoid robot company UBTech is offering up to $18 million a year to hire a chief scientist to lead its AI and robotics research. The company sold 1,079 human-sized robots last year, bringing in $119 million — 41% of its total revenue — and is racing to scale up production. The offer reportedly matches salary levels at OpenAI and Meta, highlighting how intensely China's robotics industry is competing for top talent.
TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)
🎯 Build a savings goal tracker with progress bars and deposit history
Build a savings goal tracker. Let me create multiple goals with a name, target amount, current savings, deadline, and color. Show each goal with a progress bar, percentage complete, amount remaining, and how much per month I need to save to hit my deadline. Include a deposit input to add money to each goal. Show an overall summary at the top. Use a dark lime green color scheme. React and Babel, all in one HTML file.What this does:
Seeing all your goals at once changes how you save. Each goal shows the exact monthly amount you need to stay on track. The “Add $” button on each goal lets you log deposits and watch the progress bar grow. Having a Japan Trip goal sitting at 40% is weirdly more motivating than having $1,800 sitting in a generic savings account.
What this looks like:
WHERE WE STAND(based on today’s news)
✅ AI Can Now: Review your federal tax return and catch deductions that a human preparer missed — including one case where a professional understated savings by 67%.
❌ Still Can't: File your taxes for you. AI tax tools prepare and review returns, but you're still on the hook for checking the work and submitting it yourself.
✅ AI Can Now: Power a drone through a GPS-jammed zone using only LiDAR and inertial sensors, with no cameras or satellite signals required.
❌ Still Can't: Replace all the tiny electrical components needed to build AI data centers. Transformers and circuit breakers made overseas are still bottlenecking half of the US data center projects slated to open this year.
FROM THE WEB
RECOMMENDED LISTENING/READING/WATCHING
Ray Kurzweil argues that the neocortex runs on a single repeating algorithm, and that once we understand it well enough, we can replicate it in silicon. He builds the case methodically, drawing on neuroscience, pattern recognition research, and his own work in AI, and by the end he's making claims about consciousness and machine intelligence that are either visionary or wildly optimistic depending on your tolerance for Kurzweil. He's been wrong on timelines before and right on others, which makes him hard to dismiss outright. Worth reading less as prophecy and more as a window into how the people actually building this technology think about what they're doing.
Your AI is resolving tickets. Is it keeping customers?
Resolution rates look great. But Gladly's 2026 Customer Expectations Report reveals the metric most CIOs are missing — and what the data says about where AI investments actually translate into retention, not just throughput.
Thank you for reading. We’re all beginners in something. With that in mind, your questions and feedback are always welcome and I read every single email!
-James
By the way, this is the link if you liked the content and want to share with a friend.
Some * designated product links may be affiliate or referral links. As an Amazon Associate, I earn from qualifying purchases. This helps support the newsletter at no extra cost to you and Amazon makes a tiny hair less.
