An Engineer's Log
@an_engineer_log

Founder @Agent37HQ (YC F26) | ex @Amazon @CarnegieMellon

 San Francisco
Based in United States
agent37.com
Joined December 2024
127 Photos and videos
After 15 applications and over 9 years of applying, I finally got into YC as a solo founder. full story below 👇
146
17
576
108,400
Coding in the cloud (literally) ☁️
9
42
1,152
haha, my man Tony!
1
1
71
You can now use QM for free on Agent37. Most AI agent harnesses are single-player. YCombinator's QM harness on the other hand was built for teams. Every teammate gets their own agent with their own computer to work on their tasks. Link Below👇
15
4
57
4,496
Love the voice
1
1
17
thank you :)
1
9
I love nyc
12
1
66
4,335
as you should
1
45
Love this marketing idea!! Free coffee just outside of yc @inkbox_ai @getcontextdev
10
2
63
4,500
True ! And now also free re-posts and engagement like this :)
4
119
this is my first ever X post! hi, i'm shirley! i'm building AnyAlpha, backed by @ycombinator. AnyAlpha is an agentic trading platform, it turns your idea into agents that execute for you 24/7. you get wall street data + wall street trading infra! i worked in different kinds of fintech: trading infra at big banks -> trading terminal at LSEG -> crypto trading infra at coinbase would love to chat with fintech founders, traders & sports traders! dm me and let's grab coffee. super grateful to @dessaigne and @nemild 🧡, and to @garrytan @amiklas @kul for the best group office hours 🧡🧡
315
34
1,440
85,208
Welcome Shirley !!
2
446
Replying to @an_engineer_log
curious how well the agents coordinate when tasks start depending on each other
1
1
20
It's actually just 1 sandbox per person/team member. The rest is still similar to that 1 person using a regular harness. Makes it easier to deploy to a team
1
15
Replying to @an_engineer_log
Most agent harnesses are one person at a desk; QM gives each teammate a separate desk and computer. That makes parallel work easier, with coordination becoming the next challenge.
1
2
20
yes
18
Replying to @an_engineer_log
So is this meaningfully different from giving each person a Grok bot or a Claude computer-use session, or is it mostly the same model + tools with better team plumbing? Curious where the real separation is, persistent per-user sandboxes, permission boundaries, Slack-native rooms, or just orchestration. Also: for anyone who’s self-hosted or tried the hosted version, how locked down are the sandboxes in practice? What can an agent actually reach?
2
2
53
The harness is largely similar, it's just that the multi user and a sandbox per user is build into the QM harness
1
1
60
It is unfortunately a race to the bottom.. but maybe that's ok, if the world has trillions of agents
Ivan Burazin
@ivanburazin
Oct 2
Marketing your product as the cheapest solution will get you a decent no of customers initially. There's only one problem: someone will always be cheaper. The primary moat should always be something else.
6
32
1,063
yeah but how are new models going to be funded if margins are razor thin
1
1
9
frontier intelligence can still charge a premium
1
11
Replying to @an_engineer_log
This is a game-changer for team-based AI workflows. Giving every teammate their own agent and computer makes collaboration much more scalable. 🔥
1
1
26
Yes, QM is great, and built for teams
26
Replying to @an_engineer_log
Looking good!
1
1
36
thank you Jeremy
1
29
agent37.com/qm

Agent37 - Host any agent

Run Hermes, OpenClaw, Claude Code, or Codex around the clock. Deploy yours in one click from $3.99/mo, or create agents for your users with one API call.

agent37.com
1
160
Marketing your product as the cheapest solution will get you a decent no of customers initially. There's only one problem: someone will always be cheaper. The primary moat should always be something else.
23
1
63
13,069
haha, kinda agree with this. But it is a wedge to build a better product and gain market share.
1
155
if anyone wants something shipped for context, just dm me and there's a good chance I get it out in <24 hours
Jake Casto
@0x15f
Oct 1
Replying to @mynameisyahia
You’re solving my use cases faster than I can try using Context for them
3
17
1,236
more replies
what? lmao
1
38
just impressed with the 24 hr SLA !
2
18
Replying to @JinjingLiang
wait really ? I haven't felt it
1
1
875
We raised a $4.5M seed to let agents text people. Since April: 50K+ developers, 450K monthly npm downloads and 10x revenue. Bring your agent to iMessage, WhatsApp and more → tryphoton.ai
148
80
847
1,060,065
congrats guys!
251
Excited to be co-hosting this with @insforge and @MonidHQ Take your shot at a trip to Vegas !
Tony Chang ☁️
@tonychang430
Oct 1
We are cohosting the “Build an Agent” hackathon on Oct 7 with @MonidHQ and @Agent37HQ! The challenge: build an agent that takes over a repetitive workflow you’d happily never do again. The grand prize includes $3,700 cash, a year of Agent37, and a trip to Vegas 👀 Got a workflow you’re tired of doing manually? Come build with us :) Excited to be cohosting with yall! @shengkunye @feiyou_guo @Jasperli0122 @an_engineer_log Apply Below ⬇️
1
3
13
690
Load more