Jerry Chen (@jerrychen) header
Jerry Chen (@jerrychen) profile picture

Restless. Irreverent. Greylock GP. @GreylockVC

San Francisco, CA
Joined March 2008
Jerry Chen retweeted
Where will durable value land in agentic coding: the model, the harness, or the gateway? Join Greylock Partner @jerrychen as he moderates a discussion with leaders from Microsoft AI, Nous Research, and OpenRouter at Microsoft for Startups Innovation Basecamp on November 18 in San Francisco. Register by November 12: lnkd.in/gW2hHsXB Free for founders. No Ignite pass required.
1
1
1,204
Jerry Chen retweeted
Congratulations to @SevenAICyber, @cogent_security, @CylakeAI, @Netic_AI, @parallax_ind, and @Tenzai_Labs on being named to The Information's 50 Most Promising Startups of 2026. 🎉 Proud to partner with these exceptional founders and teams from day one!
3
4
10
2,132
Jerry Chen retweeted
Ok but actually our vibe coded booth looks sick — come check us out at @aiDotEngineer CODE in SF this week and see a demo
This is what I get for vibe coding our booth design
1
1
17
1,101
Jerry Chen retweeted
This is what I get for vibe coding our booth design
8
1
1
30
1,737
Jerry Chen retweeted
just ask your agents what they prefer, slack or ando?
2
7
941
Jerry Chen retweeted
Announcing Dagger for GitLab. Run the same checks before and after push. See results in your terminal and merge request. Cut out the middleman: just GitLab and Dagger Cloud. Stop wasting compute: checks are cached, and you pay for exact usage, not machines. dagger.io/blog/dagger-for-gi…
1
10
1
34
3,536
Jerry Chen retweeted
Altara's early team reminds me of some of the strongest startups I've ever backed: super-strong, super-technical.
Some of Altara’s early technical team. I’m grateful to be building every day alongside incredible teammates who bring deep expertise at the intersection of AI and scientific research.
1
4
34
4,191
Jerry Chen retweeted
product development is increasingly about consuming information, understanding tradeoffs, and making judgments
4
2
22
3,358
Some of Altara’s early technical team. I’m grateful to be building every day alongside incredible teammates who bring deep expertise at the intersection of AI and scientific research.
7
12
2
66
8,436
Jerry Chen retweeted
On my way to SF for @aiDotEngineer CODE, to give a talk and host the first ever Niteshift conference booth. Booth demos were a huge part of Datadog’s early GTM success (we’re an ex-Datadog founding team), because we did things a bit differently: - Engineer staff: devs working directly on the product were the primary booth staff - Demos, demos, demos — the whole goal of the booth was to give as many (great!) demos as possible. At one AWS re:Invent, I remember us giving 20k-30k demos out of a total of 60k for the conference in just a few days - Demos are visually distinctive from 30 feet away and looked like nothing else on the market. Screens were oriented vertically. - 5-8 minutes per demo, tightly scripted. Though since the engineers knew the product, we frequently went off script - No chairs. Staff were always looking up and engaged, greeting passerby and looking eager to demo (yes, shifts were very hard work) - Extended follow up questions handed off to an “expert” desk in the center of the booth to keep demos moving - No one too senior to demo. Even as a VP managing hundreds of engineers (with a full schedule of customer meetings), I made time to demo every AWS re:Invent - Integration logos: biggest objection was always “ok sick demo, but no way this will work with *my* stack” So, we had an integrations page with a wall of logos of every DB, queue, language, framework, cloud service out there (along with the actual integrations to back it up) - Swag: everyone loved the purple Datadog t shirts. Insanely soft, so even if tech t shirts weren’t your vibe, they were great for the gym, as pajamas, or for giving to your partner. I still own about a dozen of these shirts. At @niteshiftdev, we have to chart our own path, but the core principles remain. We’ve worked hard on our first set of booth demos and I have a suitcase full of the first ever NITESHIFT t-shirts. We are excited to share it with you all. Hit us up if you’ll be at the conference!
5
2
1
31
1,849
Jerry Chen retweeted
Transitioning dev to the cloud is as inevitable as moving production to the cloud, but it’s going to take time, and we (ie @niteshiftdev and other providers) are going to have to win users over by simply shipping a better product. Can’t expect users to move for any other reason.
And I will say: what I do these days or chose not to do, has less to do with laziness or taste but just finding something that works and stick with it until it no longer does. I have made multiple attempts now with using cloud dev environments of various choices, including different sandboxing solutions and I just keep coming back to using my own infrastructure which I already have.
1
1
6
820
Looks like the evolution of work is going from BYOC (bring your own computer/phone) to BYOA (bring your own agents). So who's paying for the tokens? 😂
6
14
1,756
Jerry Chen retweeted
Agentic AI is moving from pilot to production. What does the stack need? Microsoft AI, Nous Research and OpenRouter weigh in, moderated by Greylock's @jerrychen at Ignite. Register: msft.it/6014aAQyi
6
9
1
48
7,692
Jerry Chen retweeted
Niteshift has always been cloud-only, which makes us a bit different than other tools out there. Because of this focus, we’ve been able to handle the most complex environments and accelerate the work of top teams with cloud agents
Conductor Cloud users looking for alternatives have joined teams like @appliedcompute, @ListenLabs, and @ambrook in using Niteshift. Many of you missed having a macOS app, so today we're launching Niteshift Desktop! We're also launching Campaigns for long horizon automations.
5
2
19
1,019
Jerry Chen retweeted
Conductor Cloud users looking for alternatives have joined teams like @appliedcompute, @ListenLabs, and @ambrook in using Niteshift. Many of you missed having a macOS app, so today we're launching Niteshift Desktop! We're also launching Campaigns for long horizon automations.
19
5
4
48
4,604
Jerry Chen retweeted
Niteshift is magically ergonomic. It just works. Sets up your dev environment, spins up previews and verifies its own work. The engineering done by the Niteshift team is incredible and I can't wait for you to experience it.
Conductor Cloud users looking for alternatives have joined teams like @appliedcompute, @ListenLabs, and @ambrook in using Niteshift. Many of you missed having a macOS app, so today we're launching Niteshift Desktop! We're also launching Campaigns for long horizon automations.
4
3
30
1,486
Quanton engine is a game changer for all @ApacheSpark users!
⚡ Quanton engine is really freakin fast. Here’s the receipt. Uber put out a blog that unpacks Quanton evaluation at production scale against OSS Spark and found it 4.7x-5.5x faster. xitter.twitterapi.xyz/UberEng/status/2118409…
6
609
Jerry Chen retweeted
More than 200,000 CPU cores power Uber’s data lake ingestion. At that scale, every efficiency gain counts. Making better use of that compute means examining how Spark executes queries, moves data, and writes to growing tables. Accelerating one stage can leave other bottlenecks untouched. Uber engineers are improving production workloads with native vectorized execution and evaluating new ways to accelerate write-heavy pipelines. Explore the engineering choices behind those efforts, where the benchmarks showed gains, and what remains to be optimized.
1
9
3
57
6,448
Jerry Chen retweeted
⚡ Quanton engine is really freakin fast. Here’s the receipt. Uber put out a blog that unpacks Quanton evaluation at production scale against OSS Spark and found it 4.7x-5.5x faster. xitter.twitterapi.xyz/UberEng/status/2118409…
More than 200,000 CPU cores power Uber’s data lake ingestion. At that scale, every efficiency gain counts. Making better use of that compute means examining how Spark executes queries, moves data, and writes to growing tables. Accelerating one stage can leave other bottlenecks untouched. Uber engineers are improving production workloads with native vectorized execution and evaluating new ways to accelerate write-heavy pipelines. Explore the engineering choices behind those efforts, where the benchmarks showed gains, and what remains to be optimized.
2
4
1
18
1,414