AI by Aakash
Everything you need to know about AI. Nothing more.
Something satisfying happens in offices, studios, and home desks every Thursday morning. People open one email, read for fifteen minutes, and walk into their next meeting knowing more about what happened in AI that week than anyone else in the room.
That email is this newsletter.
220,000+ subscribers and growing fast. Engineers, designers, marketers, founders, product leaders, and operators across every industry where AI is changing how work gets done. Which, at this point, is every industry.
They subscribe because this newsletter does something most AI coverage doesn’t: it respects their time and trusts their intelligence.
I don’t aggregate headlines. I don’t rewrite press releases. I don’t publish “top 10 AI tools” listicles. I don’t assume you need AI explained to you. I assume you’re smart, you’re busy, and you need someone doing the homework so you can make better decisions faster.
Every week, I pick the one thing that matters most, build with it hands-on, pull it apart, and write the guide I wish someone had written for me. Then I cover everything else worth knowing in a single page.
One email. Fifteen minutes. You walk away knowing what happened, what it means, and what to do about it.
What You Get Every Week
Every issue has two parts: the deep dive and the weekly update. Same structure every time. You open it and know exactly where to find what you need.
Part 1: The Deep Dive
This is the centerpiece. One topic, covered completely.
I spend the week installing the tool, running the workflows, hitting the walls, reading the documentation nobody reads, and writing up what I find. The goal is simple: by the time you finish the deep dive, you should be able to use the thing or understand the shift well enough to make real decisions about it.
These aren’t summaries. They aren’t hot takes written from a press release. They’re the kind of guide where you open it on Monday, follow the steps, and have a working system by Tuesday. Whether you’re an engineer optimizing a pipeline, a designer exploring generative tools, a marketer automating creative testing, or a founder deciding what to build next.
Three types of deep dives rotate through the newsletter:
Hands-On Tool Guides
I go from zero to expert mode so you can skip the learning curve.
The Nano Banana 2 guide walks you from your first image generation prompt to advanced techniques that most users never discover. I tested dozens of prompt patterns, documented which ones actually produce good results, and organized everything by skill level so you can start wherever you are. Designers use this to understand what’s possible. Marketers use it to generate campaign assets. Engineers use it to understand the model’s capabilities and limits.
The Complete Guide to NotebookLM covers everything: how to use it, where it breaks, when to reach for something else, and the specific workflows that make it worth opening over ChatGPT for research tasks. Anyone who works with large documents, reports, or research will find their workflow changed after reading this.
The Claude Cowork guide is the resource I wish existed when I first started running a desktop AI agent. I’ve been using Cowork daily since launch, and the guide covers setup, real workflows that save time, and the mental model shift required to work effectively with an agent that controls your computer.
The Codex guide explains why Codex quietly became the most useful thing inside ChatGPT, and shows you exactly how to use it for research, daily tasks, and structured work that used to take hours.
These guides get bookmarked. Readers come back to them weeks later when they’re ready to go deeper or when a colleague asks “how do I use this?”
Builder Playbooks
I open-source the systems I actually use. No gatekeeping. No “DM me for the template.”
My Claude Code setup with 5 free skills is the exact configuration, skills, and workflow I use to produce content every single day. You can copy the files into your own setup and start using them immediately. Over 100 iterations went into these skills before I published them.
The Claude Skills Tutorial teaches you how to build your own skills from scratch. How SKILL.md works, how to structure skills for reliability, and how to make them self-improving so they get better over time without manual rewrites.
Steal 6 of My Claude Skills gives you six more production-grade skills you can drop into your setup today. Each one was built through dozens of iterations, tested on real outputs, and refined until it worked consistently.
Claude Code v2.1 covers the update that changed how I work, with specific instructions on what’s new and how to take advantage of features most people haven’t explored yet.
When I say “open-source,” I mean it. The files are there. The instructions are there. You can be running the same system I use within an hour of reading the guide.
Trend Analysis With Teeth
When something big happens in AI, most newsletters tell you what happened. I tell you what it means and what to do about it.
When Karpathy dropped autoresearch and 42,000 people starred it on GitHub in the first week, every article focused on the machine learning angle. My guide spent two weeks pulling apart the repo, the community forks, and the real-world applications people were building on top of it. The core insight everyone missed: the loop works on ad copy, email sequences, video scripts, job posts, landing pages, design systems, and anything else where you can define what “good” looks like. I ran it on my own content and showed the scores going from 41% to 92% in four rounds. You don’t need a GPU. You don’t need to know anything about ML. If you produce anything with AI, this pattern will change how you think about quality.
When Gemini 3 launched, I had early access. Everyone else wrote about the model benchmarks. I wrote about the infrastructure play underneath it that’s rewriting the economics of how AI gets built and deployed. If you’re choosing which models to build on, which APIs to commit to, or which platforms to trust with your workflows, this is the analysis that matters.
When DeepSeek came back matching frontier model performance at a fraction of the cost, I broke down exactly what it means for pricing, competition, and which products are about to get disrupted.
When the next data centers started going to space, I covered why the physics and economics are starting to make sense and what it signals about where the AI infrastructure buildout is heading.
These are the pieces people forward to their teams and Slack channels. They’re designed to be the single resource you need on a topic, so you never have to stitch together information from five different blog posts and three Twitter threads.
Part 2: The Weekly Update
After the deep dive, every issue covers the full week in AI. This is your complete briefing.
Top story with analysis. The biggest news of the week, with real context on why it matters for people who build things. When Google shipped its biggest Maps upgrade in a decade, I didn’t just describe the features. I explained why embedding Gemini into products that 2 billion people already use is a distribution play that no other AI company can match right now. That analysis matters whether you’re an engineer choosing platforms, a designer watching the tools shift, a marketer planning campaigns, or a founder deciding where to compete.
The other news that mattered. Every other announcement, launch, and development worth knowing, each in a tight paragraph. Travis Kalanick secretly building a robotics company across 110 cities. A tech entrepreneur using ChatGPT to design a custom mRNA cancer vaccine for his dog. An engineer in Morocco open-sourcing a phased array radar that replaces $250,000 defense contractor hardware. You get the story, the context, and the link. No fluff.
Resources. The best new tools, features, and tutorials I found that week. Claude’s new inline visualizations. Self-improving skills. Subagent support in Codex. Only things I’d actually recommend to a friend.
Tools. Product launches worth paying attention to. Google Stitch going after Figma. Claude Code getting remote control. The tools section tracks what’s shipping, not what’s being announced.
Fundraising. The raises that signal where money is moving. Cursor raising at a $50 billion valuation. xAI poaching Cursor’s product leaders. The fundraising section tells you what bets the market is making and why.
Around the Web. The best AI content from other creators that week. Interviews, threads, videos, and essays I think are worth your time. This is where I point you to the other voices worth following.
You can scan the weekly update in five minutes or read it fully in ten. Either way, you’re caught up.
What a Typical Issue Looks Like
You open the email Thursday morning. Here’s what you see.
The intro is two paragraphs. It tells you what the deep dive covers this week and why it matters. No lengthy preamble.
Then the deep dive. This week it might be a complete guide to a new AI tool, a breakdown of a technique that changes how you work, or an analysis of a market shift that affects what you’re building. It’s 2,000 to 4,000 words. It has screenshots, code blocks, infographics, and step-by-step instructions where they’re needed. It reads like the smartest person on your team explaining something over coffee, not like a textbook.
After the deep dive, there’s a sponsor section. One sponsor per issue, always a tool I’ve personally vetted and actually use. No programmatic ads. No banner noise.
Then the weekly update. Top story with analysis. The other news in quick paragraphs. Resources, tools, fundraising, and around the web.
The whole thing takes about fifteen minutes if you read every word. Most readers tell me they read the deep dive carefully and scan the update for what’s relevant to them. Either approach works.
You close the email knowing exactly what happened in AI that week and what, if anything, you should do about it.
What Readers Say
“I used to have 30+ tabs open every morning trying to stay current on AI. Now I just read this on Thursdays and I’m more informed than I was before.”
“The Claude Code setup guide saved me weeks of trial and error. I copied the skills, ran them the same day, and my workflow changed overnight.”
“I forward the deep dives to my entire team. We’ve started using them as the basis for how we evaluate AI tools before adopting them.”
“Most AI newsletters feel like they’re written by someone who read a press release. This one feels like it’s written by someone who actually used the product for a week before writing about it. Because it is.”
“The autoresearch guide turned into a two-hour team discussion about how we evaluate our own AI outputs. That’s the kind of thinking this newsletter generates.”
“I’m a designer and I thought this would be too engineering-heavy. It’s the opposite. The tool guides helped me understand what’s worth learning and what’s hype.”
“As a marketing lead, the autoresearch piece alone changed how I think about testing copy and creative at scale. I sent it to every copywriter on my team.”
Who Reads This
This is a specific newsletter for people who make things with technology. The content reflects that.
Engineers and developers who are building with AI models, APIs, and tools every day and want the context that helps them make better technical decisions. Which model to use when. Which tools are worth integrating. Where the platforms are headed. The deep dives give you the analysis that documentation doesn’t.
Designers who are watching AI reshape their tools, workflows, and job descriptions in real time. Generative image models. AI-native design tools like Stitch going after Figma. The deep dives cover what’s actually usable now, what’s coming, and how to stay ahead of the shift instead of being surprised by it.
Marketers and growth professionals who need to understand AI well enough to use it, not just talk about it. The autoresearch loop for optimizing ad copy. AI image generation for campaign assets. Automated testing at a scale that was impossible two years ago. The deep dives give you the systems and the tools section gives you what’s new.
Founders and operators building AI-native products or integrating AI into existing ones. The weekly updates track where the market is moving, who’s raising, and which tools are shipping. The deep dives give you frameworks you can apply to your own product and business decisions.
Product leaders who sit at the intersection of engineering, design, and business, and need to understand AI across all three. The deep dives give you the technical context without assuming you write code, and the strategic context without assuming you don’t.
Executives and leaders who need to stay informed without spending their week on research. Fifteen minutes every Thursday keeps you current enough to lead conversations about AI, ask the right questions, and make decisions without relying on secondhand summaries from your team.
Subscribers work at Google, Meta, Amazon, Apple, Stripe, Shopify, Snowflake, Netflix, Spotify, and hundreds of startups building with AI right now. They range from individual contributors to C-suite executives. What they share is a need to stay current on AI without it becoming a second job.
Who Writes This
I’m Aakash Gupta.
I build with AI tools every morning before I write about them. That’s the whole editorial standard. When I write about Claude Cowork, I’ve been running it daily for months. When I write about Karpathy’s autoresearch, I’ve already run it on my own content and I show you the round-by-round improvement scores. When I write about Codex, I’ve built workflows with it that week.
I open-sourced my entire production system. The CLAUDE.md. The skills. The workflow templates. Readers can see exactly how I work and copy whatever is useful to them. Over 100 iterations went into the skills before I shared them.
Before going full-time as a creator, I spent a decade in growth and product roles at Affirm, Apollo, and thredUP. I’ve shipped products used by millions of people. I have an MBA from Wharton. I was a national champion policy debater, which probably explains why my deep dives are structured the way they are and why the analysis has the edge it does.
I went full-time because I realized I could reach more people by writing than by building at one company. That bet has played out. 290K followers on LinkedIn. 197K on X. 220,000+ newsletter subscribers. And the audience is growing faster now than at any point in the last three years, because AI is no longer a niche topic. It’s the topic.
The Platform
AI by Aakash is where I cover AI tools, trends, and how to use them. You’re looking at it right now.
Product Growth is my original newsletter, read by 220,000+ subscribers, covering product strategy, career growth, and how the best teams operate. I’ve been writing it for years and it’s become one of the largest product newsletters in the world.
The Growth Podcast features long-form conversations with senior leaders about how they build, hire, and make decisions. Recent guests include Dave Killeen (CPO at Pendo), Lisa Huang (SVP Product at Xero), and Christian Kleinerman (EVP Product at Snowflake). The interviews go deep on specific decisions, trade-offs, and frameworks that don’t show up in blog posts. Listen on Spotify, Apple Podcasts, or YouTube.
Land a PM Job is my cohort-based program for people breaking into product management. Three cohorts and counting. 75 seats per cohort. Live sessions, mock interviews, resume feedback from trained coaches, and a community of people actively job searching together.
290K followers on LinkedIn. 197K on X. I write on both platforms daily about AI, product, and building in public. The posts are often the first draft of ideas that become full deep dives in the newsletter.
Free vs. Paid
The weekly AI Update with the deep dive and full news roundup is currently free. You get the complete guide, the weekly update, the tools, the resources. Nothing is held back from the free experience.
Product Growth has a paid tier with extended versions of the deep dives, additional use cases beyond what the free issue covers, career content, and the full archive of every guide I’ve published.
Common Questions
How often do you publish? Every Thursday. One issue per week. No random mid-week sends, no “bonus” emails cluttering your inbox. You can plan around it.
Is this too technical for me? If you can use ChatGPT, you can follow the deep dives. They start from the beginning and build up. I write for smart people who are busy, not for people who already know everything. Jargon gets explained when it shows up. Steps are numbered when they matter. Screenshots are included when they help. Designers and marketers read this alongside engineers and they get equal value from it.
Is this too basic for me? Probably not. The deep dives go well past surface level. Senior engineers and technical founders regularly tell me they learn something new from every issue. The tool guides go deep enough to cover edge cases and failure modes. The trend analysis pieces make arguments that experienced builders find worth debating. If you’re already an expert on the week’s specific topic, the weekly update alone keeps you current on everything else happening in AI.
I already read other AI newsletters. Why this one? Most AI newsletters are written by people who read about AI. This one is written by someone who builds with AI every morning and documents what works. The deep dives are tested firsthand. The recommendations come from actual use, not from scanning Product Hunt or summarizing a launch blog. If you’ve been reading AI newsletters that feel like rewritten press releases, this will feel different from the first issue.
What’s the difference between AI by Aakash and Product Growth? AI by Aakash covers AI tools, trends, and how to use them. It’s for anyone who works with technology. Product Growth covers product management strategy, career growth, and how the best teams build. It’s more focused on people in product roles. Some readers subscribe to both. Many start with one and add the other later. If you’re primarily interested in AI, start here.
Can I read past issues before subscribing? Yes. The full archive is at aibyaakash.com/archive. Pick any deep dive and see if it matches what you’re looking for. I’d recommend starting with whichever topic is most relevant to your work right now.
Do you accept sponsors? One sponsor per issue, always a tool I’ve personally vetted and use. No programmatic ads. No banner ads. No affiliate links disguised as recommendations. The sponsor section is clearly labeled and separated from the editorial content. Readers tell me the sponsors are often as useful as the content itself, because I only work with products I’d recommend anyway.
I want to sponsor the newsletter. How? Reach out at aakash@aakashg.com. I’ll share the media kit with audience demographics, open rates, and pricing. Sponsorship slots often book weeks in advance.
Your Thursday Morning, After You Subscribe
You wake up. The email is in your inbox. You open it with your coffee.
Fifteen minutes later, you know what the most important AI development of the week is and how to use it. You know what else happened and whether any of it affects your work. You have a list of tools worth trying and a sense of where the market is heading.
You walk into your first meeting with more context than anyone else in the room. When someone asks “did you see the Karpathy thing?” or “what’s the deal with the new Gemini?” or “should we be looking at this tool?” you have a real answer. Not a vague sense that you saw a headline somewhere. A real, informed, specific answer.
Your designer mentions an AI tool they’re excited about. You already read the deep dive and can tell them exactly where it works and where it falls apart. Your engineer asks about a new model. You know the benchmarks and, more useful, you know the strategic context behind who built it and why. Your CEO asks what’s happening in AI this week. You have the full picture in fifteen minutes of reading.
That’s what this newsletter does. One email. Every Thursday. Everything you need. Nothing you don’t.
Subscribe now. When the first email arrives, drag it to your primary inbox so it lands where you’ll see it. Thursday mornings, your entire AI briefing is waiting.
