Inspired by KOLs to Follow on X — Tech Builders and Creators and the original X thread.
Following more people on X is easy. Following the right people is harder.
If you work in AI/ML engineering, solutions architecture, or technology product building, your timeline should not only be a place to catch the latest news. It should become a radar system: a way to notice new tools, new patterns, how founders think about products, and how strong engineers explain trade-offs.
This list is not trying to collect as many accounts as possible. It is a deliberate shortlist: who to follow first, what to learn from them, and which group fits which learning goal.
1. Choose people who upgrade your judgment
An account is not worth following only because it is famous. For builders, a good account usually helps in one of four ways:
- It helps you understand technology from first principles — especially AI, LLMs, infrastructure, databases, and runtimes.
- It helps you see how products are built — from founders, operators, DevRel, and design engineers.
- It helps you catch early signals — new tools, new workflows, and changes in the ecosystem.
- It improves your professional taste — how to write, decide, reason about trade-offs, and evaluate quality.
If your time is limited, do not follow from FOMO. Follow according to the role you want to strengthen.
2. A shortlist to start with
If you only pick a few accounts first, start here.
| Account | Follow if you want to |
|---|---|
| @karpathy | Understand LLMs and AI from first principles, not just at the tool layer. |
| @bcherny | Track deep perspectives on Claude Code and AI coding workflows. |
| @trq212 | Catch fast updates around Claude Code and how the community is using it. |
| @kepano | Learn about PKM, Obsidian, tool philosophy, and durable product design. |
| @addyosmani | Follow production-level patterns in AI and web engineering. |
| @leerob | Track signals around Cursor, Next.js, DevRel, and AI developer tooling. |
| @rauchg | Understand how Vercel thinks about infrastructure, deployment, and AI-native web apps. |
| @GergelyOrosz | Stay updated on engineering leadership, the tech market, and how software teams operate. |
| @mitchellh | Learn systems thinking from the person behind HashiCorp and Ghostty. |
Think of this as a core radar: AI, developer tooling, product taste, and engineering judgment in one place.
3. AI, LLMs, and dev tools
This group is useful if you build with AI, use AI coding tools every day, or want to understand where developer tooling is going.
- @karpathy: best for people who want to understand AI from the foundations, especially LLMs. Read slowly, not only as a feed item.
- @addyosmani: useful if you care about web performance, AI product patterns, and bringing AI into real products.
- @bcherny: worth following if Claude Code is part of your workflow, or if you want to understand AI coding tools from a builder’s perspective.
- @trq212: good for quick updates and practical tips around Claude Code.
- @leerob: useful for builders tracking Cursor, Next.js, the Vercel ecosystem, and how DevRel tells product stories.
- @ctatedev: helpful if you want to see experiments from Vercel Labs and AI devtooling.
The shared value of this group is that they show AI as more than models. AI is changing IDEs, deployment, frameworks, workflows, and how developers learn their craft.
4. Founders and operators for product judgment
A strong builder does not only know how to code. You also need to learn how to choose problems, ship products, operate teams, and say no.
- @rauchg: follow for platform thinking, developer experience, infrastructure, and AI app deployment.
- @mitchellh: good for people who care about system design, terminals, infrastructure, and how technical founders think about craft.
- @dhh: read as a strong source of opinions on frameworks, business, operations, and independent product building.
- @jasonfried: useful for company building, simplicity, and avoiding unnecessary complexity at work.
- @karrisaarinen: worth following if you care about product taste, productivity tooling, and how Linear builds product experience.
- @Shpigford: useful for makers and founders who want to learn from the process of building many small products.
This group keeps your timeline from becoming tool-only. You will see more about decision-making: why build this, why ignore that, and how to design a team and product around a clear point of view.
5. Design engineering for better product taste
If you only follow backend, infrastructure, and AI accounts, it is easy to miss interface quality. But AI-native products still need good UI: clear flows, smooth micro-interactions, understandable state, and flexible design systems.
- @shadcn: useful if you build modern web apps and care about component quality.
- @emilkowalski: good for learning animation, transitions, and micro-interactions.
- @joshpuckett: useful if you want to sharpen your interface craft.
- @raphaelsalaja: helpful if you like UI pattern analysis.
- @nandafyi: follow for a design perspective inside a highly technical product environment like Cloudflare.
- @mengto: useful if you are learning design while building products.
- @jh3yy: strong for demos, breakdowns, and interesting frontend details.
This group is especially valuable for AI builders. When many products use similar models and frameworks, the difference often shows up in workflow, interaction, and polish.
6. Engineering media for industry context
Not every account needs to teach you code directly. Some accounts help you understand context: what the industry is discussing, how companies are changing, and what engineers are debating.
- @GergelyOrosz: strong on engineering leadership, hiring, the tech market, and inside stories from software companies.
- @theo: useful if you follow web development, framework debates, and hot takes from the frontend/fullstack community.
- @ThePrimeagen: good for low-level thinking, editor workflows, performance, and reaction content.
- @Rasmic: useful for web development tutorials and accessible content.
- @atmoio: another source of engineering content to widen your radar.
Treat this group as industry sensing, not as a source of truth. Read to notice what the community is paying attention to, then verify through practice.
7. Database founders for backend and architecture
AI apps still need a strong data layer. Databases, backend architecture, and edge architecture matter even more when products need realtime behavior, global distribution, or better developer experience.
- @jamwt: follow Convex from a CEO perspective, especially if you care about reactive backends.
- @jamesacowling: useful for deeper technical perspectives around Convex and backend architecture.
- @glcst: good if you care about Turso, edge SQLite, and databases moving closer to users.
- @samlambert: worth following if you want to understand PlanetScale, serverless MySQL, and database operations.
If you work in solutions architecture, do not skip this group. They help you see the database not only as a place to store data, but as part of the product architecture.
8. Frontier labs to keep up with new model releases
If you work directly with LLMs, following people at each lab lets you see product direction earlier, instead of only reading news that has already been filtered through someone else.
- Anthropic: @karpathy, @bcherny, @trq212 — already listed in the AI/LLM/DevTools group above; they are also among the closest voices to where Claude is heading.
- OpenAI: @polynoamial (reasoning research, lots of technical detail), @gabriel1 (Sora), @jxnlco (dev experience, shares a lot about Codex).
- Google AI: @OfficialLoganK (official Gemini/AI Studio updates), @ammaar (product and design, vibe-coding in AI Studio), @fofrAI (creative use cases for generative models).
- Cursor: @leerob (already listed above), @ericzakariasson (insights on using Cursor), @mntruell (Cursor’s CEO, major releases).
- xAI: @milichab, @skcd42 (Grok updates), @ai_explorer25 (broader AI content and free resources).
This group does not replace the AI/LLM group above — it adds an insider view from each lab, so you notice how Claude, GPT, Gemini, and Grok are diverging before it shows up in aggregated news.
9. Turn your timeline into a learning system
Following the right people is only the first step. The more important question is how you use the timeline.
A simple system:
- Create a list for AI / LLM / DevTools.
- Create a list for Design Engineering.
- Create a list for Founder / Operator / Product.
- Create a list for Frontier Labs (Anthropic, OpenAI, Google AI, Cursor, xAI) to track model updates close to the source.
- Each week, save 3–5 posts into your knowledge base.
- For each saved post, write one line: “Where does this change how I build or think?”
If an account only makes you feel busy but does not help you think better, mute or unfollow it. A good timeline is not the one with the most information. A good timeline sharpens your judgment.
10. Takeaway
Your follow list should reflect the professional version of yourself you are building.
If you want to become stronger at AI engineering, start with Karpathy, Addy Osmani, Claude Code accounts, and AI dev tooling builders. If you want to build better products, add founders, operators, and design engineers. If you want to understand the industry, follow engineering media. If you work on backend or architecture, do not ignore database founders. If you need to keep pace with new model releases, add the frontier-lab group for each lab.
Follow fewer people, but with more intent. When your timeline becomes a learning radar, scrolling X can become a daily upgrade loop — not just another content habit.