Evan Reiser

For the 3rd year in a row, @abnormalai is a top 20 company on the @Forbes annual Cloud 100 list! Thank you to Abnormal AI team + customers + partners! More futuristic products & AI stuff coming soon to help protect our customers from the future of cybercrime.

@GeoffreyHuntley “Now that you have seen all my prompts since the beginning how would you rearchitect the project”

this is my favourite prompt of all time: “how could this be better?” reply with yours and why it rocks!

@MetamateDaz Try building something for fun in Claude code for like an hour

how do you get your work ethic back? I burned myself out from 2018-2025 and now I’m tired and just don’t want to anymore.

@signulll Small tam joke Esp if short

i recently met someone named “tam” & i made a total addressable market joke. let’s just say it didn’t size well.

We're also launching Detection 360 Insights, Custom AI Models (which define a detection pattern from a single example), and AI Phishing Coach with BEC/VEC simulations built on real org relationships. More here: abnormal.ai/blog/attune-em…

Attune already powers 85% of our detections, catching 150K more attack campaigns per week with 50% higher precision. It caught a novel Teams OAuth phishing campaign through calendar invites two months before researchers documented the technique.

Security spent 30 years memorizing bad. Bad IPs, bad domains, bad signatures. AI attackers just broke that model. Every attack is now unique, no shared infrastructure, no reused payloads. The only durable signal left is whether someone is behaving normally.

I've been a sci-fi nerd my whole life. I love everything about it. The way a good story takes something you think you understand and makes you see it completely differently. 2026 is the year I realized I get to live inside one. Things that were plot devices three years ago are shipping as products. AI cloning voices from a five-second clip. 52 million Americans spending over an hour a day talking to AI companions. Agents spawning copies of themselves across the internet without anyone asking them to. The world is moving so fast that I started doing something I've never done before: writing science fiction. Each piece follows the same pattern. Take something happening in AI right now that seems minor or innocuous. Extrapolate it forward 10-20 years to see what implicit future we're heading toward. Then bring it back to the present to reframe how we look at that seemingly small thing today. The series is called Implicit Futures. The first published piece is AI Value Injection. AI shapes how your workforce thinks by 0.04% per day. Imperceptible. Compounded over four years, that's 100%. I followed it forward twenty years and wrote what I found.

How well someone thinks they use AI for their job is inversely correlated with how well they actually do. The more you dig in the more you realize how far there is to go. If AI makes you feel like a beginner every week that's probably a good sign.

Despite 100x gains in productivity with AI agents There will be *more* security engineers in the future not less

@ShrivuShankar If AI makes our async work 10x more production then there are certainly many meetings that are no longer worth doing relative to the opportunity cost

ai native meeting tip: before any meeting, run your agenda through claude (connected to your docs via glean/atlassian/mcp, or even just a local md archive) and have it generate strawman proposals for every decision then change the meeting from "discuss X" to "debate claudes take on X" way easier to poke holes in a prewritten memo than to state-transfer verbally in real time. once the context is wired up, claude can often just one-shot quite a bit of it sometimes its good enough you can just cancel the meeting

@ShrivuShankar I think Claude is more forgiving of dictation typos than humans are

the whisperflow-to-claude pipeline is creating a new dialect. its not how you talk to people and its not how you write, its this third thing where you just think out loud in fragments and trust the other side to assemble it ive had a few awkward moments on slack and zoom where i forgot the other side wasnt wisper-claude

me getting ready for board meeting after anthropic is doing $1B NARR/week

@alexiskold Abnormal.ai

What are examples of small, vertical focused models trained from scratch that aren’t based on OpenAI or Claude and work really well?

The AI conversation is stuck on "faster, cheaper, more efficient." Future models will hold your entire company in their head and connect dots humans physically can't. The intelligence compounds every year. A billion ants can't write Shakespeare.

@ShrivuShankar Try the visual explainer skill with this!

ok decompiling things with opus is actually very addicting here's apparently how netflix and slack work - gist.github.com/sshh12/dda3a89… - gist.github.com/sshh12/4cca8d6… idk maybe wont be that hard to vibe code these after all > use chrome dev tools to explore and provided a grounded tear down analysis of how the site works

@lulumeservey Are you sure you are reading this right? Maybe it’s a hostage video?

Going direct only works when it’s authentic Really kills the mood if it feels like the CEO’s blinking in Morse code So much of this is forced: 1) fake off the bat, with the exaggerated surprise at the size of the burger and presence of onions. Are we to believe he has never tested or even seen this burger before? 2) “I love this product” 🙃 should’ve just gone ahead and called it a SKU 3) the fries propped up facing outward adds to the staged vibe. We know they’re McDonald’s fries, we don’t need to see the properly oriented logo, be chill 4) the ™️ after “Big Arch” in the captions is further giving corpo slop 5) overhyping the size (“I don’t even know how to attack it”) backfires to make it seem underwhelming. It’s respectable but not in the top quartile of burgers by size; this is America 6) “patties, sauce, some lettuce — there’s so much going on with this burger” OK but for the record, that is approximately the bare minimum going on for something to qualify as a burger The only part that felt unscripted was him using the Lord’s name in vain while pretending to be awed by the size of the burger. Unnecessary; take that to Ben and Jerry’s

@romanbuildsaas > What if building at full speed… is the reward? This is the way!!! 💪🚀

The “problem” with AI agents is that I don’t enjoy doing anything anymore except working. It feels like I’m inside a massive video game, and my character is leveling up at an exponential rate. In 6 months, we’ve compressed what should have taken 4 years. I wake up at 9am and I’m at my computer until 11pm. I force myself to go to the gym. I force myself to be social. But the only thing that truly excites me is progressing in the game. Maybe it’s just a phase. Isn’t the goal to build, exit, and then finally enjoy life? But what if this is enjoying life? What if building at full speed… is the reward?