Moti Tabulo (@MTabulo) profile picture

đŸ€– AI-native product builder ,🚀 Entrepreneur đŸ„·đŸ» ex-Product @ Immunefi, Code security @ Klevoya

Joined July 2014
Photo by Moti Tabulo (@MTabulo)Photo by Moti Tabulo (@MTabulo)Photo by Moti Tabulo (@MTabulo)Photo by Moti Tabulo (@MTabulo)Photo by Moti Tabulo (@MTabulo)
Moti Tabulo retweeted
My biggest takeaways from Rob Snyder, author of The Power of Pull: 1. Solving a pain point isn’t enough to get people to buy your product. Rob’s first startup checked all the classic boxes: hair-on-fire problem, a differentiated solution, and design partners using it daily. Yet he still spent two years pivoting aimlessly. Every prospect agreed it solved real problems, but none bought. 2. Rob’s "PULL" framework identifies four conditions that drive a buying decision: The buyer has a Project on their to-do list, and a reason it’s Unavoidable right now. They also have a List of options they’ve considered, each with Limitations that leave them stuck. That person is a “coiled spring,” desperate to find a product to buy. 3. The right cold outreach message needs to be interesting, not persuasive. Its only job is to get someone to take the call. Rob calls this “founder magic” outreach. One startup’s entire message was: “Hey, Snoop Dogg invested in us, and I thought you’d find it interesting too. Would you be open to chat?” No pain points, no value propositions, no personalization. Just something that gets a conversation. Rob’s LinkedIn benchmarks are 20 requests a day, 30% to 40% accepted, 30% to 50% of those replying, and most saying yes. 4. Design partners are a trap. Nearly every time, the partner buys into your vision, hands you a pile of feedback, and never buys. Then you realize “all of their feedback was just like them playing Steve Jobs.” The people who have something important to get done don’t want to be design partners. They just want it done. Offering a partnership is “a negative signal to people who actually have pull, and a positive signal to people who don’t.” 5. Pull belongs to a specific person, not to a company. A common founder mistake is to say “we sell to SpaceX” or “we sell to mid-market private equity firms.” In practice, you’re selling to the head of legal at SpaceX, who has a particular project on her to-do list she can’t solve with her current team. The company doesn’t have pull. She does. 6. The signal of real pull in a meeting isn’t just enthusiasm—it’s someone leaning in and asking how to move forward. “Wow, that’s amazing, totally the future of our industry, I can see how that would save us a ton of money” is not pull. That’s a polite exit. What Rob looks for is a “pull-shaped action”: “Wait, you can do that? How do we get started? Does the product exist yet?” Before the product even existed, someone in pharma told one founder, “If you can do that, I will pay you right now.” 7. Run the first sales call like a detective, not a pitcher. Rob’s opening question is always some version of “I’m sure you get a lot of people reaching out. Why did you take this call?” The best-case outcome: they fill out the PULL framework for you unprompted. They tell you exactly what they’re trying to do, why now, and what they’ve already tried. 8. When a call goes lukewarm, throw the Hail Mary. Ask, “Is there anything on your to-do list that you would pay me to do right now?” A team selling applicant tracking software to governments got a lukewarm reaction but then learned the buyer would pay for help with job descriptions. That became their first contract. When founders are truly stuck, Rob recommends going into “goblin mode,” a burst of slightly crazy effort to get in front of customers, like the founder whose parents went door to door to dental offices for him. 9. Rob calls "The Case Study Factory" his most important idea. The job is to find one type of customer who says 'hell yes' before and after buying, then repeat that success through pipeline, sales, and delivery. This customer forms your repeatable case study. A startup finds one repeatable case study and then repeats it a million times. To grow as fast as possible, borrow the theory of constraints from manufacturing: find the bottleneck in the process of repeating that case study and treat it as the only thing that matters until it’s fixed.
An 86-minute guide to finding product-market fit Rob Snyder (@RobSnyder1) is the author of The Power of Pull, has helped over 50 startups go from zero to $1M+ ARR, and is the most requested guest by my podcast listeners. We discuss: 🔾 The PULL framework for finding a customer who’s "weird not to buy" 🔾 Where the best startup ideas come from 🔾 The cold outreach message that gets calls 🔾 How to run your first sales call (and why he skips design partners) 🔾 The "Hail Mary" question and "goblin mode" Listen now 👇 youtu.be/19qrLchbQEc
109
64
14
995
89,968
Moti Tabulo retweeted
THIS IS ACTUAL F*CKING GOLD Anthropic just shared one of the best tricks for Claude Code. The idea: run Sonnet, Haiku and Opus 5.5 together, each one doing only the job it's best at. Opus-level judgment on every big call, without paying Opus rates for every routine turn. You just launch Claude Code like this: claude --model sonnet --advisor opus here's the split: → Sonnet 5.5 runs the session and writes the code → Haiku 5.5 explores files and docs in parallel → Opus 5.5 stays on call as advisor for the big calls → you stop spending Opus on work that doesn't need it Opus only steps in at three moments: → before a plan locks → when the same error shows up twice → before calling it done Sonnet builds. Haiku explores. Opus reviews. JEV engineering is the same move one layer down: the forks that need no thinker (which file, which tool, retry or stop) go to JEV in milliseconds, and the big models only see the ones that split the full tree > Sonnet 5.5 runs the main session and edits code > search maps the repo on Haiku 5.5 > grep finds call sites on Haiku 5.5 > docs pulls the API reference on Haiku 5.5 > Opus 5.5 on call as advisor, reads the whole transcript when consulted > JEV routes micro-forks in milliseconds paste the tree and this prompt into Claude Code ↓ "Rebuild my Claude Code setup around this tree: 1. Run the main session on Sonnet 5.5 with Opus 5.5 as advisor. 2. Create three read-only Haiku 5.5 subagents: search (maps the repo), grep (finds call sites), docs (pulls the API reference). 3. Add one rule to CLAUDE.md: consult the advisor before locking a plan, after the same error twice, and before calling a task done. 4. Route fast deterministic forks (tool choice, path selection, retries) through JEV. Show me every change as a diff first. No edits until I say go." ↳ code.claude.com/docs/en/advi
 save it, you'll use this every single day.
REA is pure f*cking treasure right now it lets your AI agent reverse engineer any app, down to the binary Opus 5.5 reads how a feature works, then rebuilds its own version from scratch but one wrong number can pass every check in this article :
60
123
6
1,632
193,771
Ben Affleck being an AI-pilled founder was not on my bingo card for 2026. Did you know Netflix bought his AI startup for $586M in March? đŸ€Ż
Ben Affleck reveals he writes Python, understands convolutional neural networks, worked extensively with GPUs, and used his celebrity status to get private looks at Google and OpenAI’s video models “I’ve always been kind of into computers since I was young. Then, when film started to move from analog film to digital, I became more interested in that aspect of it. The visual-effects workflow for many years has included machine learning, so I can write pretty shitty Python scripts and stuff like that. “With convolutional neural networks, which were the precursors to what the transformer can do, which is much more computation simultaneously, you would do things like look at what’s called a tensor. That’s the numerical translation of a visual image in numbers, like the batch number, the frame number and the red, green and blue values of each pixel in each frame. It’s just that simple. That numeric is called a tensor. “You’d use a convolutional neural network to identify patterns that reveal what’s called edge detection or feature extraction, which is identifying patterns well enough to know, this is where the window ledge is, so we can more easily take the green-screen image out and replace it with something. “That was familiar to me early on because, prior to Artists Equity, I had a small visual-effects company. I’ve worked with GPUs a lot too. The visual-effects guys said, ‘Hey, you should see. There are a couple: Google and this other company, OpenAI, are doing really interesting stuff with transformers in video.’ “I’ve learned that I can actually just call up and go, ‘Hey, it’s Ben Affleck. Can I come see what you’re doing?’ Sometimes people say yes, to my astonishment.”
1
3
214
Love this Jev explanation!
Holy crap. Rick and Morty just explained Jev AI to me better than any tech demo could.
3
88
Moti Tabulo retweeted
It’s easy to dismiss Jev it as “just a classifier”. But people (me included) who have been training encoder-style models for classification for many years know they were usually special-purpose and limited in some way. The breakthrough of Jev is that it generalizes well. And I’d say the secret sauce is probably more in the data than in the training algorithm. (Plus a nice API design on top of it.)
poor guy claim to have built Jev a year ago but no one cared, and now Jev stole all the thunder many people are saying “you gotta tell your story” or “marketing is important”, and they just completely missed what actually made the difference here i just looked into this laya model laya.convaiinnovations.com/ and: - it only supports 512-1k context
 a lot of use cases won’t fit at all - evaluating the model directly shows its accuracy is as good as a coin flip. in order to get good results, you need to first fine tune it i’m sorry, but that’s not Jev there’s a massive gap between an interesting research and a useful product you can “tell your story” all you like, but you can’t blame Jev for stealing your thunder when Jev did all the work to make a well-packaged solution anyone can just grab and go Jev is not completely new from an academic sense, just like how ChatGPT was not the first LLM don’t underestimate the effort and value in putting together something that’s actually good enough for adoption - it makes all the difference
135
205
26
2,178
242,559
I’m 45. Solo builder from Cambridge, UK. Looking to connect with other builders and AI enthusiasts!
8
1
16
497
I tagged all 421 replies to this across 358 accounts. 108 want more automated review. 93 ration human attention by risk. 63 replaced inspection with evidence. 33 review the plan before the diff.
If you’ve stopped doing human code reviews: how is it going? If you’re still doing them: how are you able to keep up with the explosion of code + PRs, and how do you prioritise what to review “properly”?
8
19
310
I went through all 400+ replies to Gergely Orosz's question about reviewing code that agents wrote. 350+ accounts. The argument I was expecting is humans versus AI, but that's not actually what's happening. In fact, the unit of review is changing.
1
104
I went through all 400+ replies to Gergely Orosz's question about reviewing code that agents wrote. 350+ accounts. The argument I was expecting is humans versus AI, but that's not actually what's happening. In fact, the unit of review is changing.
1
1
3
289
Those leaning on AI reviewers have their own problem: AI reviewers are too willing to find things, and noise becomes its own bottleneck. The most-liked reply in all responses is @jacobtechtavern, on a crusade against AI review comments. Their fix is another agent filtering them.
2
27
The real question per change is how much assurance it needs. @abhirb already routes on it: low-risk auto-merges, security-sensitive gets a human. Code isn't the scarce resource anymore, it's confidence. So the future of code review? Spend human attention only where it matters.
2
24