Alfred Lin

@adamshuaib conviction = intuition + courage

Smart move by Nominal. Fid Labs' agents automate complex HW eng and robotics tasks. Now they'll be even stronger sitting on top of Nominal's HW data platform. Let's go.

Last night, Harvey signed our 101st Am Law 200 firm, so we now serve the majority of Am Law 200 and 60+ of the Am Law 100 firms. Job's Not Finished.

@Deepak_AvairAI Agreed it is currently on leadership to push it top down

@Alfred_Lin Adoption is a leadership problem, not a technology problem. Every company I've seen succeed with AI had a CEO who used it daily before mandating it for anyone else.

Inside the tech industry, AI is all anyone can talk about. Outside, we are still incredibly early in adoption. Even the divide between those using AI at all vs using it daily is stark. Leaders need to drive usage from the top, and support employee adoption by setting expectations, even if imprecise. Good enough is better than perfect for now. As Dan put it in the below email to his company: “The one place you absolutely cannot be is zero.”

@jonathanmendez @sequoia @gdb @OpenAI All thanks go to Greg

@Alfred_Lin @sequoia @gdb @OpenAI This was a fantastic conversation. Thanks for sharing.

At @sequoia’s AI Ascent last week, @gdb told me something that stuck: in late 2024, AI wrote ~20% of @OpenAI's code. That number is now 80%. We also got into why human attention, not compute, is the real bottleneck in AI-augmented work, plus what it might mean to run an org of 100,000 agents. We’re grateful to Greg for joining us and many of the top founders/builders in AI. You can watch the full video here: 00:00 Intro 00:49 Compute Hunger Explained 02:13 Scaling Laws Mystery 03:31 New Architectures Ahead 04:42 How Close to AGI 06:46 Startup Playbook for AI 09:24 Inside OpenAI with Codex 11:11 Teams and Governance Shift 14:52 Security and Responsible Deployment 25:33 Science Frontiers and Wrap Up

@tizimmer Agreed. It's important leaders give permission to learn unabated.

@Alfred_Lin funny how most "resistance" disappears once leadership makes it socially safe to be bad at it first

@MurtazaAhmed_AI Indeed

@Alfred_Lin The best is when work starts to feel strange without it.

Incredible to watch Stripe still ship and innovate like it's Day 1.

Sir @demishassabis has a mind for synthesis. His favorite book is about a grand theory of everything. His preferred philosophers are seen by some as opposites. His life's work ranges from board games to Nobel-winning science. We're grateful to have hosted Demis and his @GoogleDeepMind team at @sequoia AI Ascent last week for a fireside chat. He kindly gave us permission to share this, and you can watch the full video here: 00:00 Intro 00:38 The Common Thread 01:29 Games as AI Training 02:59 Startup Advice 1.0 04:39 Founding DeepMind 07:25 DeepMind and AGI 08:52 AI for Science 10:37 Biology Breakthroughs and Isomorphic 12:42 New Sciences 20:29 Philosophy

@karpathy and I are back! At @sequoia AI Ascent 2026. And a lot has changed. Last year, he coined “vibe coding”. This year, he’s never felt more behind as a programmer. The big shift: vibe coding raised the floor. Agentic engineering raises the ceiling. We talk about what it means to build seriously in the agent era. Not just moving faster. Building new things, with new tools, while preserving the parts that still require human taste, judgment, and understanding.

@AdamDudleyX Fully agree. Engineering is an easy place to drive fast adoption to start.

@Alfred_Lin Smart, and I think this should be expanded to all other areas of a business as well, which the email touched on. Otherwise, you end up with a huge usage and leverage gap between engineering and the rest of the functions, which is likely the state of things at most companies.

.@chainguard_dev has set a new expectation for their engineering leaders for AI adoption: every leader should be around the 50th percentile of Claude Code token usage among their direct reports. Why 50th? <50th percentile: - Not enough firsthand experience to know what’s possible - Can’t scope work/expectations accurately - Can’t coach their teams on tools appropriately >50th percentile: - Have found edge of what’s possible - Need to be the multiplier, not the power user - Enable team instead of doing it yourself The gap between teams getting massive leverage and those barely touching the tools may naturally resolve itself over time, but companies can’t afford to wait.

@mstockton This recent post from @jack and @roelofbotha gets into some of the role/org items you're mentioning: x.com/jack/status/20…

Very much liked this article. I don’t think token utilization is the right metric, but it’s directionally correct, and might be the best proxy we have right now for ‘are you doing the AI thing or not?’ Also agree, and this has always been true, that a manager has to find the right balance between truly knowing how to do the work and not actually doing the work. It’s a complete failure if you are at either end of the spectrum. I am not convinced that we are to simply adopt old org structures for these new times though. I’d much rather have 3-5 absolutely astounding AI first builders than a team with layers where you actually have to think about things like who does what and mentorship, etc etc - but this obv still doesn’t scale far, even with AI. Pipe dream I guess. One take-away that is undeniable though. The only way to do this is to actually do this. Wherever you are, you need to be trying things daily and updating your priors

@geetharganesh Yes, this should be work related. See: Accountability section.

@Alfred_Lin Does it matter whether the leaders consume tokens for work-related initiatives that move the needle for the org or work on side projects ?

I've talked about the importance of outcomes over output (x.com/Alfred_Lin/sta…). But in lieu of waiting for perfect metrics, anything is still better to drive early adoption. From Dan's note: "Everyone has to pay the tuition to learn these tools. The cost of falling behind is a lot higher than the cost of tokens."

@Alfred_Lin This is a huge liability for any startup. You can have a clear plan and a good judgment, still be slow to deliver at scale due to token consumption, aside any other security pitfalls. Tokens consumption is the new engineering debt in AI era.

@immad @mercury Congrats, Immad and team. Huge accomplishment.

1/ Today @Mercury received conditional approval from the OCC to establish Mercury Bank, N.A. I started Mercury in 2017 to build the bank I wish had existed as a founder. Nearly a decade later, we’re getting there. 🧵

@andrew__reed You've entered your Cool Dad phase: "During its Top 10 heyday in the late 90s and early oughts, Andrew was one of the "cooler" classic boy names. While there are still plenty of babies named Andrew, it now feels more like a Dad Name than baby name."

Somehow, this is not a joke

@bhalligan Sage advice

No matter what kind of company you are...start making your internal company data legible to AI. Today. As a founder, you are essentially building two versions of your company: the one humans work in and the digital twin that AI agents navigate to do the heavy lifting for you.