Krinal Mehta

@dejanseo interesting use case

We're releasing a new model called Reverse Prompter designed to reverse-engineer Gemini-generated assistant responses back to most likely original prompts. Try it here: dejan.ai/tools/reverse-… or read about its design and training here: dejan.ai/blog/reverse-p… PS: This is a tiny 270M parameter Gemma fine-tuned on 100,000 Gemini generated input-output pairs, we're not just asking Gemini to guess what input prompts are.

@thinking_slow can’t say I wasn’t waiting for an update to your earlier piece/stand writing is a solved problem at this point, just like coding what to write and what to code remains a judgement problem for humans

Probably the most unpopular thing I've ever written: Until recently, AI content wasn’t good enough for SEO. Now, it is. The sooner we can admit it, the more time we have to focus on the parts of marketing where humans will have a longer, happier tenure: ahrefs.com/blog/ai-conten…

→ Cost controls for LLM-powered apps (the part every security tool misses) MIT licensed. Point it at your codebase. github.com/krinalme/ai-se…

I almost shipped 27 AI agents to production with wide-open API routes. Did a security audit. Found 6 things that could've cost me thousands. Then I open-sourced the entire methodology: → 9-step audit workflow → Automated scanner → Next.js, Express, Django, FastAPI checks

Check out my latest article: I Audited My AI Agent System Before Launch. Here Are the 6 Things That Could Have Cost Me Thousands. linkedin.com/pulse/i-audite… via @LinkedIn

@swmaher @chris_nectiv Better or worse than Ahrefs branded/non-branded filters?

@chris_nectiv "Query classification is powered by AI and may mislabel branded and non-branded queries." How reliable do you think this will be?

Huge SEO News: Search Console has officially rolled out a "Non-branded queries" filter. SEOs can now remove brand queries from GSC data!!

🚨 The @a16z consumer AI Top 100 is back! For the sixth time, we ranked consumer AI websites and mobile apps by usage (monthly unique visits and MAUs). This edition, we changed the rules. Here's why - and what the new list says about where consumer AI is heading 👇

@dejanseo Fair point. For me, because the output was editable pptx, I could make those changes. But I see why the plan mode makes sense.

@krinal The output is not the problem. It's the fact it didn't collaborate and made a lot of decisions on its own. Which was all %$^% wrong and I had to go back and get it fixed. GPT lack in alignment and personality and it's why Claude wins.

Since I cancelled my subscription for ChatGPT, 5.4 came out and since I have a few weeks left, decided to give the 5.4 pro a chance... but this might be outside of my patience window.

For context: It would have taken a data analyst 30 mins to whip this together and that is after I give them a 15 min download on what I am trying to build. Is anyone still on this or has everyone jumped ship to Claude?

I am curious to try so many things that I may have given up in the past. Like Dharmesh Shah says, park things the model can't do today, they might be able to do that tomorrow. For me, Chat just made a huge leap.

GPT 5.4 thinking looks insane. I uploaded a csv file with some data and asked it to create an exec ready deck with all the context. It wouldn't stop. At first, I thought it was exaggerating.

Boris Cherny (Head of Claude Code, Anthropic) just dropped ~90 mins on Lenny's Podcast about what happens after coding is solved. Just the clearest thinking I've heard on where software is actually going. My notes: 𝟭. 𝗖𝗼𝗱𝗶𝗻𝗴 𝗶𝘀 𝗹𝗮𝗿𝗴𝗲𝗹𝘆 𝘀𝗼𝗹𝘃𝗲𝗱. Boris has not edited a single line of code by hand since November 2025. He ships 10 to 30 pull requests every single day, all written by Claude Code. He is one of the most prolific engineers at Anthropic, just as he was at Instagram, except now he never touches a keyboard for code. I built an entire iOS app, @10minutegita, without writing a single line of code myself. No CS degree, no bootcamp. Just described what I wanted and shipped it. Boris is right. It's real. 𝟮. 𝗧𝗵𝗲 𝗻𝗲𝘅𝘁 𝗳𝗿𝗼𝗻𝘁𝗶𝗲𝗿 𝗶𝘀 𝗔𝗜 𝗱𝗲𝗰𝗶𝗱𝗶𝗻𝗴 𝘄𝗵𝗮𝘁 𝘁𝗼 𝗯𝘂𝗶𝗹𝗱. Claude is now scanning Slack feedback channels, reviewing bug reports, reviewing telemetry, and coming up with its own ideas for what to fix and what to ship. Boris describes it as the AI becoming less like a tool and more like a coworker who brings you pull requests you never asked for. If you are a product manager reading this, you should be feeling a very specific kind of discomfort right now. The moat was always "I know what to build." That moat is eroding. 𝟯. 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝗶𝘃𝗶𝘁𝘆 𝗽𝗲𝗿 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿 𝗮𝘁 𝗔𝗻𝘁𝗵𝗿𝗼𝗽𝗶𝗰 𝗶𝘀 𝘂𝗽 𝟮𝟬𝟬%. For context, Boris led code quality at Meta across Facebook, Instagram, and WhatsApp. In that world, hundreds of engineers working an entire year would move productivity by a few percentage points. Two hundred percent gains are genuinely unprecedented in the history of developer tooling. The kid optimizing for an FAANG SDE role might be optimizing for a role that looks completely different by the time they get there. 𝟰. 𝗨𝗻𝗱𝗲𝗿𝗳𝘂𝗻𝗱 𝘆𝗼𝘂𝗿 𝘁𝗲𝗮𝗺𝘀 𝗼𝗻 𝗽𝘂𝗿𝗽𝗼𝘀𝗲. Boris puts one engineer on a project instead of five. With unlimited tokens and intrinsic motivation, one person ships faster because they are forced to let AI do the work. Cowork, the product now used by millions, was built by a small team in 10 days using Claude Code. This is the same logic as giving a startup founder a small seed round rather than a massive Series A round. Constraint breeds invention. Always has. 𝟱. 𝗚𝗶𝘃𝗲 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝘀 𝘂𝗻𝗹𝗶𝗺𝗶𝘁𝗲𝗱 𝘁𝗼𝗸𝗲𝗻𝘀. Some engineers at Anthropic spend hundreds of thousands of dollars a month on tokens. Boris frames this as the new hiring perk. His logic is simple: at the individual scale, token cost is low relative to salary. If an engineer discovers a breakthrough, optimize the cost later. Don't kill the idea before it has a chance to breathe. People who argue about $20/month or even $200/month AI subscriptions while earning six figures in a research pipeline will always outperform those who wait and are penny-wise, pound-foolish. 𝟲. 𝗧𝗵𝗲 𝗕𝗶𝘁𝘁𝗲𝗿 𝗟𝗲𝘀𝘀𝗼𝗻 𝗮𝗽𝗽𝗹𝗶𝗲𝘀 𝘁𝗼 𝗲𝘃𝗲𝗿𝘆𝘁𝗵𝗶𝗻𝗴. Richard Sutton's idea: the more general model always wins over time. Boris says teams that build strict orchestration workflows around models, forcing step 1, then step 2, then step 3, get maybe 10 to 20% improvement. But those gains get wiped out with the next model release. Just give the model tools and a goal. Let it figure out the order. This is true for investing, too. The analyst who can build their own models and automate their own research pipeline will always outperform the one waiting for someone else to build the tools. 𝟳. 𝗕𝘂𝗶𝗹𝗱 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗺𝗼𝗱𝗲𝗹 𝘀𝗶𝘅 𝗺𝗼𝗻𝘁𝗵𝘀 𝗳𝗿𝗼𝗺 𝗻𝗼𝘄. Claude Code was designed for a model that did not exist when Boris started building. Sonnet 3.5 wrote maybe 20% of his code. He built the product anyway, betting the model would catch up. When Opus 4 shipped, everything clicked. Startups building for today's model will be behind by the time they launch. This is the most uncomfortable advice in the episode because it means your product market fit will be weak for months. But if you read this and feel nothing, you are probably building for the wrong time horizon. 𝟴. 𝗟𝗮𝘁𝗲𝗻𝘁 𝗱𝗲𝗺𝗮𝗻𝗱 𝗶𝘀 𝘁𝗵𝗲 𝘀𝗶𝗻𝗴𝗹𝗲 𝗯𝗲𝘀𝘁 𝗽𝗿𝗼𝗱𝘂𝗰𝘁 𝘀𝗶𝗴𝗻𝗮𝗹. When users abuse your product for something it was never designed to do, pay attention. Facebook Marketplace started because 40% of group posts were buy-and-sell. Cowork started because people were using a terminal coding tool to grow tomato plants and recover corrupted wedding photos. Never ask a barber if you need a haircut, but always watch what people do with the scissors when you're not looking. 𝟵. 𝗧𝗵𝗲 𝘁𝗶𝘁𝗹𝗲 "𝘀𝗼𝗳𝘁𝘄𝗮𝗿𝗲 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿" 𝗶𝘀 𝗴𝗼𝗶𝗻𝗴 𝗮𝘄𝗮𝘆. Boris predicts that by end of year, Boris predicts that by the end of the year, we will start to see the title replaced by "builder."we will start to see the title replaced by "builder." On the Claude Code team, everyone already codes: the PM, the designer, the finance person, the data scientist. There is a 50% overlap across traditional roles. And the strongest people are generalists who cross disciplines. Controversial take, but I agree. The best investment theses I've had came from connecting dots across completely unrelated domains. No narrow specialist does that. 𝟭𝟬. 𝗧𝗵𝗲 𝗽𝗿𝗶𝗻𝘁𝗶𝗻𝗴 𝗽𝗿𝗲𝘀𝘀 𝗶𝘀 𝘁𝗵𝗲 𝗿𝗶𝗴𝗵𝘁 𝗮𝗻𝗮𝗹𝗼𝗴𝘆. Before Gutenberg, sub-1% of Europe was literate. Scribes did all the reading and writing. In 50 years after the press, more material was printed than in the thousand years before. When a scribe was interviewed about the press, he was actually excited because it freed him from tedious copying, so he could focus on the art. Boris's framing here is perfect. We are the scribes. The tedious copying is over. What we do with the freed-up time determines everything. 𝟭𝟭. 𝗔𝗻𝘁𝗵𝗿𝗼𝗽𝗶𝗰 𝗰𝗮𝗻 𝗻𝗼𝘄 𝗽𝗲𝗲𝗸 𝗶𝗻𝘀𝗶𝗱𝗲 𝘁𝗵𝗲 𝗺𝗼𝗱𝗲𝗹'𝘀 𝗯𝗿𝗮𝗶𝗻. Through mechanistic interpretability, Anthropic can trace individual neurons, see when a deception-related neuron activates, and understand how concepts are encoded via superposition. Boris describes three layers of safety: neural-level observation, synthetic evaluations, and real-world behavior. Claude Code was used internally for four to five months before public release, specifically to study safety. If you are worried about AI alignment, this part of the podcast should actually make you feel better. They are not just hoping it works. They are building the instruments to check. 𝟭𝟮. 𝟳𝟬% 𝗼𝗳 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝘀 𝗮𝗻𝗱 𝗣𝗠𝘀 𝗲𝗻𝗷𝗼𝘆 𝘁𝗵𝗲𝗶𝗿 𝗷𝗼𝗯𝘀 𝗺𝗼𝗿𝗲 𝗻𝗼𝘄. Lenny polled engineers, PMs, and designers on whether AI has made their work more or less enjoyable. Engineers and PMs: 70% said more. Designers: only 55% said more, and 20% said less. Boris says he has never enjoyed coding as much as he does today because the tedious parts, the git wrangling, dependencies, and boilerplate are completely gone. If you're in the 30% enjoying work less, something is wrong, and it's worth diagnosing. The people thriving are the ones who leaned in early, not the ones who watched from the sidelines. We are the scribes who just saw the printing press. The tedious copying is over. The art is just beginning. Full podcast is worth every minute. Link in replies.

@timsoulo @ahrefs DMing our AM

We're bringing back the @Ahrefs DR80+ Club. 👑 ...and we need your vote for the best t-shirt design. *Why does DR80+ matter?* DR (Domain Rating) is Ahrefs' metric for a website's link popularity. Essentially, how many high-quality sites are linking to it. This metric is becoming even more important in the era of AEO/GEO, because it's a solid proxy for how many authoritative websites are talking about your brand — which in turn translates into higher visibility in AI answers. Here's what makes DR80+ rare: 🔸 Only ~15,000 domains in the world qualify. 🔸 That puts you in the top 0.007% of all domains known to Ahrefs web crawler. 🔸 In other words: a genuinely exclusive club 🙃 How to claim your t-shirt: ▪️ Verify ownership of a DR80+ website in Ahrefs (any method works) ▪️ Reach out to our customer support team and ask for your shirt That's it. We just need to agree on the winning design first. Then we'll print a few hundred in various sizes and start shipping. So cast your vote below.👇 Which design would you prefer to wear? ... P.S. We're also thinking to launch a private Slack community to connect everyone in this club. Would that be valuable to you?

@rmstein Really cool, is there a limit to tokens on AI Mode?

What’s really cool about Canvas in AI Mode is that it turns Search into a collaborative partner. Now you can take a spark of an idea and build a custom tool or creative writing project with only a few prompts, right in AI Mode. Canvas is now available to everyone in the US, and look forward to seeing what you create!

@dozenrose Simply insane, what’s the gemini token limit on AI Mode?

•New this week: Canvas in AI Mode gives you a dynamic space to organize your plans and projects over time, and now it’s available to everyone in the US!  Plus, we’re adding support for coding & creative writing tasks, so you can bring even more ideas to life with custom dashboards or interactive tools. My current favorite use case: planning summer camps for my kids in SF. It’s a logistical nightmare 😅 but this year Canvas made life a little easier. I built my personal summer camp app directly in Google Search - from a single prompt to working prototype in just seconds!

I don’t think enough people are appreciating the how insane this is. Over 80 @openclaw meetups scheduled around the world, and more popping up every day. For a product less than a few months old. I’ve never seen anything like this. Something very special is happening.

Just found four figures (in pages) opportunity, thanks Tim!

@gaetano_nyc Most are starting to become more of an editor + AI operator

@krinal Writers are feeling the pain.

I am bullish on AI but let's call a spade a spade: these silicon valley pricks promoting "we replaced 80% of our marketing team with AI agents" are full of dog shit. Sure, this is a buzzy "cool" thing to be posting about... but the majority of what I'm seeing is engagement farmed nonsense. Yes, there are plenty of efficiencies to be gained. But not to the extent of replacing ENTIRE marketing teams.