Today we launch the next chapter of Gushwork.
Over the last 30 years, we saw commerce moving online -
But what actually moved online were low value retail products: books, clothing, electronics.
Anything high value, from $5,000 to $1M - a hydraulic crane, an industrial roofing contract, a heat shrink machine - still runs predominantly on human buyer seller conversations.
And conversation was the one thing software couldn't do.
That just changed.
We are bringing 35 million sellers of high value products & services in the physical economy to the agentic web -
One of the largest categories of commerce that barely moved online in the internet before AI.
one of the most challenging launch video's i've done till now,
what the team came to me with,
- pre recorded footage
- stuff already shot on a script that wasn't communicating the product well enough and lacked the best story/narrative
goals that we set to take the video to 10x :
> build a better more engaging story around their pre existing brand narrative
> use the dialogs from the existing shots and scenes to build a new story, remove repetitive dialogs.
> use ai voice overs and visuals to bridge the gap between the existing dialogs and new storyline to make it feel seamless
and after 4 days of locking in, we shipped this beauty of a video, that had everything from a storyline, to product comms, to app feature showcase, combined together to
make this.
working solo on crazy videos like this teaches me a bunch of things, and i keep getting better at how to tell good stories.
Freeport Markets (@freeportmarkets) gives retail traders buy/sell signals as breaking news happens, allowing users to trade ahead of price moves.
With $60M traded and growing 50% month-over-month, it's clear that a new generation of traders is adopting the tools once reserved for Wall Street.
Congrats on the launch @Bryan__Reed and @Lihong062.
https://t.co/wLY8dw1KmE
Freeport Markets (@freeportmarkets) gives retail traders buy/sell signals as breaking news happens, allowing users to trade ahead of price moves.
With $60M traded and growing 50% month-over-month, it's clear that a new generation of traders is adopting the tools once reserved for Wall Street.
Congrats on the launch @Bryan__Reed and @Lihong062.
https://t.co/wLY8dw1KmE
my cold email to @thetanmay pitching him short form content repurposing back in 2020 when it wasnt a thing
it might've not closed but the reply gave me the confidence to keep sending 50-100 cold emails every day as an 18 year old,
i was making money in covid lockdown when people around me (adults) were losing jobs, it was crazy and scary at the same time
shoot your shot, send that email, make that dm, trust me you'll never regret doing it, and if you don't, 1 year from now you will regret it thinking if only you hit send.
that one yes, that one reply, is only one outreach away.
and always follow up
p.s- for those who forgot, igtv used to be an instagram video thing for a little while.
(shoutout to @garyvee for that short form video style idea that i used to pitch podcasters youtubers around the world as an 18 year old )
I was building on Jupiter Predict when every dev i asked said the same thing : "the API isn't out yet"
the web app was running everything was live, and i just couldnt sit without doing something .
So mapped Jupiter Predict's production endpoints from DevTools before the API dropped
all USD values are in micro-USDC (590000 = $0.59, not documented anywhere)
Jupiter hands you a transaction to sign yourself doesn't execute it
re-serializing the message even once silently breaks the signature
Jupiter pre-signs at slot 1, you sign at slot 0 wipe slot 1 and everything breaks
built a 6-agent AI debate engine on top: Stats Analyst, Time Decay, Whale Flow, Research, Devil's Advocate + a Moderator that gives confidence %, edge over market, and Kelly-sized position
then an auto-trader that runs the full loop every 30 mins 4 gates: confidence, edge, liquidity, cooldown. if all pass → debates → decides → signs → submits → confirms on-chain
real trades. real USDC. real Solana signatures verifiable on Solscan right now.
full breakdown 👇
@sssionggg@kashdhanda@JupiterExchange@jup_predict@WillMcFletcher@weremeow@JupDevRel
i just reverse engineered @AnthropicAI's $1,000 enterprise claude security plan so you can run it on $20 claude code to fix your vibecoded projects before you ship it
Claude Security just shipped, its the final piece of security to ship product, it scans your codebase, validates vulnerabilities, patches them with minimal effort, all automated. but there's a catch. it's enterprise only and most of us can't touch it. so I went digging. turns out Claude Security runs on Opus 4.7 the same model you and I already have access to on the Pro plan. so I rebuilt 80% of it myself in 3 prompts. here's the exact pipeline:
1. scan your repo with Sonnet 4.6 (cheaper tokens): look for SQL injections, hardcoded keys, RCE, the works
2. pass findings through a filter prompt : only what qualifies as a real, severe security breach survives
3. final prompt hits Opus 4.7 generates minimal patches with file path, old line, new line, and why the fix is sufficient tested it on Juice Shop (the gold benchmark for cybersecurity) found 8 real bugs. hardcoded RSA private key, SQL injections, server-side template injection, remote code execution. every single one documented and verified.
next time you ship an AI SaaS, run these 3 prompts before you go live. dropping the exact prompts in the comments @AnthropicAI is coming for every job. might as well use it before it uses you.
The only person I've seen using OpenClaw effectively is @AlexFinn.
Using OpenClaw to build content about OpenClaw to sell a course about OpenClaw lol.
"When anyone can make anything, the big differentiator is what you choose to make."
- Paul Graham