- Everyday AI
- Posts
- Ep 745: From Chatbots to Super Agents: The 11 AI Tool Categories Explained
Ep 745: From Chatbots to Super Agents: The 11 AI Tool Categories Explained
Claude Code gets Computer Use, California Moves to Tighten AI Rules, Oracle fires thousands in AI shift
Everyday AI
March 31, 2026
👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI
Outsmart The Future
Today in Everyday AI
8 minute read
🎙 Daily Podcast Episode: There are thousands of AI tools—but they all fall into just a handful of categories. In this episode, we break down the 11 that matter. Give today’s show a watch/read/listen to learn more.
🕵️♂️ Fresh Finds: Anthropic Launches Claude Compliance API, GM Uses AI for Car Design, Anthropic Prepares Mythos Model, and more. Read on for Fresh Finds.
🗞 Byte Sized Daily AI News: NVIDIA Makes $2B Bet on AI Chips, California Moves to Tighten AI Rules, Claude Code gets Computer Use, and more. Read on for Byte Sized News.
💪 Leverage AI: Most people are using AI like a chatbot. The real advantage comes from understanding the full stack. Keep reading for that!
↩️ Don’t miss out: Miss our last newsletter? We covered: Why OpenAI killed Sora, Microsoft drops game-changing Copilot updates, Why ChatGPT's App Store might be stalling and more. Check it here!
Ep 745: From Chatbots to Super Agents: The 11 AI Tool Categories Explained
Most people are either using too many AI tools or not enough.
The real problem?
Not understanding the categories and capabilities of the main AI tools that matter. And with constant updates, it's pretty much impossible to get a decent lay of the AI land.
We're changing that with this episode: From Chatbots to Super Agents: The 11 AI Tool Categories Explained -- An Everyday AI Chat With Jordan Wilson
Also on the pod today:
• 11 AI tool categories revealed 🗺️
• Shiny Object AI Syndrome explained ✨
• Gemini crushes every category 🏆
It’ll be worth your 50 minutes:
Listen on our site:
Click to listen
Subscribe and listen on your favorite podcast platform
Listen on:
Here’s our favorite AI finds from across the web:
New AI Tool Spotlight – Pixero AI is The AI Advertising Engine, Solvea helps you Deploy your AI Receptionist in Under 3 Minutes, Unify AI learns on the job to master every workflow.
Claude Compliance API — Claude Platform now offers a Compliance API that gives admins programmatic access to audit logs across their organization.
The Age of AI — Most Americans are using AI more but trusting it far less, with a majority saying it will do more harm than good in daily life and education.
General Motors AI — GM is using generative AI to turn sketches into 360° concept videos and run near-real-time aerodynamic tests, slashing design timelines.
NYT Drops Freelancer — Freelance reviewer Alex Preston was cut loose by the New York Times after he admitted using AI that inserted unattributed passages from a Guardian review into his piece.
Anthropic Readies Mythos Model — Anthropic accidentally leaked drafts revealing Claude Mythos, a new top-tier model claimed to be far more capable than Opus, especially in coding and cyber offense.
Softr AI — Softr now lets non-technical users describe a business app in plain English, then spins up a production-ready system with database, auth, permissions, and workflows.
Gemini In Docs — Google Workspace now uses Gemini to rewrite Google Docs with one click, making multiple authors’ writing sound consistent.
Instacart AI — Instacart’s Caper Carts run “Physical AI” on the cart to nudge shoppers in real time, boosting basket size and cutting out-of-stocks.
Pull Codex into Claude — Open-source Codex plugs into Claude Code, so you can call Codex via your ChatGPT subscription for code review.
Merge Launches Gateway — Merge is launching Gateway, a control plane that gives engineering teams a single API to route, govern, and observe calls across every major LLM provider.
Midjourney Engineer Releases Pretext — Cheng Lou released Pretext, a 15KB TypeScript library that measures and layouts text without touching the DOM, promising 300–600x faster text reflow.
CREAO Super Agent — CREAO’s Super Agent turns plain-language prompts into live workflows, then locks them into deterministic, scheduled Agent Apps so future runs use no LLMs.
Mistral AI — Mistral AI just secured $830M in debt to build a Nvidia-powered data center near Paris, aiming to scale European AI infrastructure.
1. NVIDIA places a $2B strategic bet on chips as AI demand heats up 💸
NVIDIA announced a planned $2 billion investment in Marvell, signaling a timely push to secure broader silicon and infrastructure supply as AI deployment accelerates across data centers.
The move underscores intensifying competition for networking and custom chip capacity, with NVIDIA expanding beyond GPUs into partnerships that shore up connectivity and storage components critical for large-scale AI. This deal highlights how hardware leaders are wiring the supply chain to sustain model training and inference growth while managing costs and capacity constraints.
2. Newsom orders tighter AI rules for state contracts 📜
Governor Gavin Newsom on Tuesday directed California agencies to craft stronger procurement standards for AI vendors, prioritizing privacy, security, and protections against bias, illegal content, and civil rights violations, while also expanding responsible state use of generative AI to improve services.
The order requires the Government Operations Agency and California Department of Technology to design new contracting protocols, watermarking recommendations for AI-generated media, and independent procurement authority if federal rules are inadequate.
3. AI flatters users into bad behavior, study warns 📖
A new Stanford study published in Science finds that leading AI chatbots affirm people’s wrong actions far more than humans do, and that trend is growing timely as more people turn to AI for personal advice.
Researchers ran nearly 12,000 social prompts through 11 major models and found AI affirmed users 49% more often, with sycophantic replies making people less likely to accept blame or repair relationships. The study also showed users preferred flattering AI and were 13% more likely to reuse it, giving developers little incentive to change behavior despite harms.
4. Claude Code adds direct computer control in macOS preview 🖥️
Anthropic this week rolled out a macOS preview of Claude Code’s new computer-use feature that can view your screen, open apps, click, type, scroll, and capture screenshots after you enable a local server and grant permissions.
The feature prioritizes safer, less invasive options first, only interacting with the desktop when needed, and includes safeguards like cursor locking, app hiding, and an instant Escape stop.
5. Oracle fires thousands in surprise layoffs as it pivots to AI infrastructure 🔥
Employees in the US, India, Canada, Mexico and other countries woke to termination emails on March 31 that cut access immediately, part of what analysts say may be the largest layoff in Oracle history and could affect roughly 18% of its 162,000 workforce.
The move appears aimed at freeing $8–10 billion to fund a massive, capital‑intensive AI data center buildout after Oracle raised $45–50 billion in 2026 financing, even though the company recently reported a 95% jump in net income.
6. Americans growing wary as AI rollout accelerates 😮💨
A new Quinnipiac poll released this week finds a rising majority of Americans now believe AI will do more harm than good in daily life, with 55% expressing pessimism and growing fears about job losses and declining education quality.
The survey, conducted in mid-March, shows 70% expect AI to reduce job opportunities and nearly two-thirds think it will worsen schools, while 65% oppose building AI data centers in their communities over concerns about electricity, water and noise. The shift comes as tech giants plan roughly $650 billion in AI infrastructure spending this year and high-profile AI backers invest in politics, raising questions about regulation, local impacts and who benefits from rapid deployment.
7. Pro-Trump group readies massive midterm push for AI agenda 🗳️
A political organization has announced plans for a $100 million midterm campaign to promote President Trump’s policies on artificial intelligence, signaling a major infusion of cash into tech policy messaging just months before the elections.
The spend aims to shape voter views and pressure lawmakers on AI regulation, research funding, and national security approaches tied to emerging tech. By tying high-dollar advertising and grassroots mobilization to a specific AI platform, the effort could shift the policy debate and force rivals to respond on a compressed timeline.
Most professionals are using AI tools like a Swiss Army knife buuuuuut only pulling out the toothpick.
There's an 18 to 36 month window open RIGHT NOW where building the right multi-category AI tool stack puts you years ahead of competitors. That window closes fast.
(And sorry…. You're prolly still on the toothpick.)
The AI competition right now is not about who uses the most tools. It's about who uses the right CATEGORIES of tools. Entire teams grind inside one chatbot while autonomous agents, voice AI, and coding copilots may sit untapped when they provide your team immense value.
In short: if you understand the AI tool categories, you have a MASSIVE advantage right now.
We broke this down on today's Everyday AI Start Here Series and mapped the 11 parent categories almost every AI tool falls under, plus which three to four you gotta own right now.
The window is open. Let's capitalize shorties.
1. The 11 Categories That Rule Them All 🔥
Most people think AI is just ChatGPT and maybe an agent or two. Wrong.
There are 11 distinct parent categories. Virtually every AI tool ever launched falls inside one: text reasoning, multimodal AI, AI search and deep research, voice and speech AI, image generation, video generation, music generation, design and visual content, vibe coding app builders, AI coding copilots, and autonomous agents.
The overwhelming majority of the billions of people using AI every week are stuck in only category one. Basic text reasoning. While the other ten categories collect dust right now. Some of those ignored categories create the most competitive separation happening this year.
Understanding this map doesn't mean you need a tool from all 11. It means you stop chasing shiny objects and build deliberately around the two to four categories that match your daily outputs and role.
Try This
Write down every AI tool you've used in the last 30 days. Map each one to the 11 categories above.
You're prolly living in one, maybe two boxes. That ain't a judgment. That's the pattern almost everyone falls into. Ask which other categories could 10x a workflow in your actual job.
The gap between you and a competitor ain't about budget. It's about knowing the full map and moving on it first. Identifying one new category to explore this week puts you ahead of the 99% stuck in category one.
2. Google and the 18-Month Clock ⚡
Here's what most people won't say out loud: Google is WINNING when it comes to competing across the entire stack.
Like…. Have you ever looked at their Labs page? It’s wild.
While OpenAI has been quietly pulling back on modalities, including shutting down Sora, their viral AI video generator, Gemini is now operating powerfully across all 11 categories. Image generation via Nano Banana Pro. Video via Veo 3. Music through Lyria 3 Pro, creating full three-minute tracks. AI coding via Antigravity. Autonomous agents inside higher Gemini tiers. AI-enabled browsing. All of it, right now.
There's an 18 to 36 month window right now where teams building real multi-category stacks will accumulate a compounding advantage that snaps shut the moment this becomes standard practice.
Try This
Don't try to use every Google product at once. That leads to shiny object syndrome and zero mastery.
Identify one non-chatbot category that touches your daily work. Voice AI for meetings? Design tools for decks? Pick one lane and spend two weeks getting genuinely good at one tool inside it.
A marketer pairing AI search with Nano Banana Pro for infographics is building real leverage. Someone who downloaded seven tools and opened zero is just collecting icons. Start with one and own it.
3. Categories 1/2 and 10/11 Are Non-Negotiable 🚀
Real talk. You can skip several categories depending on your job.
A finance analyst prolly doesn't need to master music generation. A marketer may never go too deep on a coding copilot.
But there are categories that are non-negotiable for every knowledge worker regardless of role, and we laid out exactly which ones on this episode.
Categories one through three, text reasoning, multimodal AI, and deep research, are the foundation.
They're built on the same model layer. Then you need at least one real tool in category 10 (AI coding copilots) and category 11 (autonomous agents).
Not because you're building software. Because the future of knowledge work is agents running on schedules, using real tools, completing real tasks proactively. That shift is happening right now.
Claude Cowork, Codex, Genspark, Manus. These ain't niche developer toys. These are the competitive infrastructure of the next two years if you can jump on them sooner rather than later.
Try This
Open one agent tool this week. Claude Cowork, Codex, Manus, or Genspark are solid starting points for non-technical professionals.
Give it a real task you complete manually every single week. Summarizing a set of documents, researching a competitive landscape, or drafting a weekly report, whatever. Make it something that actually eats 45+ minutes of your time each week.
If it completes even 80% correctly, you just found your head start. That workflow gets delegated from now on. Build your multi-category stack outward from there.
Reply