Pond helps you solve any problems with AI agents and talented experts. Post a task today: joinpond.ai/ Hiring GTM and Head of Engineering. Please DM.
San Francisco
- Tweets 2,811
- Following 579
- Followers 40,373
- Likes 3,882
‼️ Past Model Factory winners, please check this message carefully👇
We need you urgently!
‼️ ATTENTION: check your email - past AI model winners on Pond!
An enterprise wants to source AI workflows through Pond, and we need your help getting them to come - there’s a sizable task posted on Pond!
I think we can win this together! Check your email and reply to me!
4
10
2,139
This morning alone: 10 under-radar YC / Speedrun acceptance announcements. And you can sell to them before everyone!
There’s a 2-week window after founders announce their offers and before the public lists. No one has reached out. No banking, no stack, no vendors. You’d be first to sell.
A fintech unicorn asked for this tool on Pond and started selling ahead of everyone. They say they’re big today because they bet early: one of their pre-seed clients became a unicorn, and they grew with them.
Want the same tool? Post a task today. In 1 day you can have it running: joinpond.ai
2
3
17
1,743
That YC / Speedrun window closes when the lists go live.
A Pond task is already pulling early signals - off-radar founders, including an ex-SpaceX Starship engineer now founding.
Don’t wait for the directory.
Post the task.
joinpond.ai
14
3
37
2,124
New task on Pond:
Warm-intro CISO and other security experts at startups to HOG3, an AI penetration testing agent that finds real attack paths.
Earn $50-$100 per intro, plus 10% of first month's revenue if it converts. No deal-closing required.
go.joinpond.ai/donely
10
3
26
2,666
This is the right way to complete future product comparison tasks on Pond:
AI agents are everywhere right now.
But with so many agents out there, the real question isn't “Which AI is the best?”
It's “Which one actually works best for the way you work?”
That's what I wanted to find out.
So, for a @JoinPond bounty, I decided to put two agents head-to-head: @SolidTheAgent and Grok @bot.
And I gave both of them the exact same prompt, the exact same requirements, and let them work completely on their own.
Here were the results 👇
6
1
23
2,465
Pond AI retweeted
One thing I’ve realized about building AI agents: Building the agent is only half the work.
The harder part is getting it into the hands of people who actually need it.
That’s why I’m excited about @JoinPond.
AI agent builders can now bring their agents to Pond, get discovered by real users, and potentially turn something they built into something they can actually monetize.
I just submitted my own: Foxy 🦊, currently under review. Coming Soon on Pond.
1
1
7
481
1 day left on the Solid vs Grok Bot comparison.
We’ve rejected submissions that used Grok instead of Grok Bot. Different product. Same name isn’t the brief.
Read the task. Complete it correctly.
go.joinpond.ai/solid-grokbot
Grok Bot users: you can win up to $200 right now.
Solid.tech is building AI workers used by people from Revolut, ElevenLabs, EY, Nvidia, and Stanford University.
Now they want your help exploring differentiators and benchmarking them against Grok Bot.
Participate and win up to $200: shorturl.at/YhUpX
Note: Please make sure you use Grok Bot (x.ai/bot) to complete the task rather than Grok
6
1
22
1,825
Pond AI retweeted
my first X post!
hi i'm lucy! asian, just moved to the bay by myself lookin for more friends in tech :)
a lil about me:
> rejected 7-fig from openai
> 3 yrs ex-google sde
> founded joinpond.ai - request AI UGC video
still new here. let's grab coffee!!
37
10
295
287,151
“Anyone know a good video editor?”
This is how you inherit 40 DMs and zero usable cuts.
The replies aren’t submissions. They’re pitches and they might deliver quality works.
Now you can post a task on Pond sourcing videos. People create many: you review and pay for results.
4
3
27
1,790
Post the task.
Get the tool.
Use the agent.
Congrats Neha @nehaaaa_6.
Starting of september....
Want to know which startups founders say “we got into YC” before the list appears?
Alpha: YC Launch Monitor - a Slack bot for the @JoinPond project that tracks YC Directory, Speedrun, and public founder posts into one alert.
DM me if you want to try it.
yc-launch-monitor-1-m0kx.onr…
7
17
1,980
1 day left:
We’re looking for 10 remote sales hustlers to work with our founder.
- Work directly with the founder and you will learn a lot.
- Access our GTM infrastructure and all resources.
- Generous commission you can make.
go.joinpond.ai/remote-sales
5
3
19
3,272
Pond AI retweeted
Headline metrics matter, but the experience of the human operator behind an agent running on agentic payment rails matters more.
Over the past 1.5 months, we ran a compensated cohort with @JoinPond. We reviewed every submission, shared feedback with the @BuildOnCircle and @arc teams, and fixed bugs during the program.
Every point of failure the cohort surfaced helped shape SELAT’s roadmap.
A few takeaways:
1. SELAT made the rails invisible.
The @JoinPond cohort was vocal whenever the UX broke. But no one mentioned “abstracting rails” or "trying to understand what x402 or MPP is" because, in practice with SELAT, the rails were already out of sight.
Neither the operator nor the agent should need to care which rail is used or which chain a merchant settles on.
2. Removing first-mile friction drives conversion and retention.
An upstream deposit bug unexpectedly created an A/B test. We worked with @BuildOnCircle to fix it mid-program.
Once funding friction was removed, most testers went on to make paid calls—and were willing to maintain $5–10 USDC balances in their wallets for agentic payments.
3. Machine-native commerce UX doesn’t usually break at the wallet layer or the payment rail.
It breaks because the core primitives required for an open market—discovery, trust, risk pricing, and recourse—are still being built.
That’s the infrastructure @selat_ai is pushing forward.
@corey__cooper @akelani @samconnerone
@ThinksDylan
SELAT is pre-launch. We go public on @arc the week of Sept 16. Before that, we ran a design-partner program on @JoinPond.
Testers installed SELAT, created a self-custody agent wallet, funded it, and paid for real APIs with real USDC. 121 submissions; 62 qualified after screening.
A few things we've learned:
SELAT makes the rails invisible. 73% of paying wallets settled over both x402 and @MPP in the same session, and nobody asked which rail they were on. The rail showed up in testers' write-ups only as a column in their own ledger.
The first mile decides everything. Testers who got through funding to a paid call rated us 7–9 out of 10 on coming back unpaid; testers who got stuck at funding rated us 2–4. Same product, same pitch. The friction was in funding. Testers surfaced it, we fixed it mid-program, and the next batch measured deposits crediting in 2–35 seconds. The ones who reached a paid call said they'd keep a balance in the wallet. Fixing the first mile is the conversion and retention lever.
Testers build on SELAT. 28 wrote their own SELAT skills from the catalog and opened PRs.
We had also shipped the transactability score without announcing it; at least one tester found it on their own and was impressed!
We don't hide imperfections. Money software earns trust by showing what broke and what changed. That's the mandate for building something with real utility.
Thanks to @JoinPond & @ThinksDylan for running this, and to everyone who tested. Launch details soon.
2
2
7
603
SELAT AI needed a design-partner cohort before launch. They posted it on Pond.
121 people applied. 62 made the cut. Real wallets, real USD, real product feedback.
Congrats @selat_ai and thanks to everyone who tested!
SELAT is pre-launch. We go public on @arc the week of Sept 16. Before that, we ran a design-partner program on @JoinPond.
Testers installed SELAT, created a self-custody agent wallet, funded it, and paid for real APIs with real USDC. 121 submissions; 62 qualified after screening.
A few things we've learned:
SELAT makes the rails invisible. 73% of paying wallets settled over both x402 and @MPP in the same session, and nobody asked which rail they were on. The rail showed up in testers' write-ups only as a column in their own ledger.
The first mile decides everything. Testers who got through funding to a paid call rated us 7–9 out of 10 on coming back unpaid; testers who got stuck at funding rated us 2–4. Same product, same pitch. The friction was in funding. Testers surfaced it, we fixed it mid-program, and the next batch measured deposits crediting in 2–35 seconds. The ones who reached a paid call said they'd keep a balance in the wallet. Fixing the first mile is the conversion and retention lever.
Testers build on SELAT. 28 wrote their own SELAT skills from the catalog and opened PRs.
We had also shipped the transactability score without announcing it; at least one tester found it on their own and was impressed!
We don't hide imperfections. Money software earns trust by showing what broke and what changed. That's the mandate for building something with real utility.
Thanks to @JoinPond & @ThinksDylan for running this, and to everyone who tested. Launch details soon.
4
2
20
1,953
This video will help you complete this task and future tasks like this:
Grok Bot users: you can win up to $200 right now.
Solid.tech is building AI workers used by people from Revolut, ElevenLabs, EY, Nvidia, and Stanford University.
Now they want your help exploring differentiators and benchmarking them against Grok Bot.
Participate and win up to $200: shorturl.at/YhUpX
Note: Please make sure you use Grok Bot (x.ai/bot) to complete the task rather than Grok
5
4
18
1,983