ANNOUNCEMENT: Rokt has just paid out over $1 BILLION to brands.
Rokt is a money printing button: Click enable and add up to 10% more profit.
RT + comment "Rokt" and I'll send you a FREE Profit Gap Agent.
Iโm Noah, the founder of Instinct.
Instinct is a personal agent that weโve been building for the past few months. The interface is simple: there are no new interfaces. You can text or call it. It's trained to use a phone and a computer in the same way that humans do. Instinct combines simplicity with extreme capability.
Iโm thrilled with everything our early users are doing with Instinct. Theyโve told us theyโve planned cross-country road trips, bought weekly groceries and concert tickets, and cancelled hundreds of dollars of subscriptions. Someoneโs even planning their wedding with Instinct.
We want to make Instinct the best personal agent for all of you. Itโs available in an invite-only beta program while weโre actively bringing up more compute. Iโm excited to see what you all do with it.
https://t.co/lVra3kd4TT
In the last month, @aftersell has had over 1 MILLION impressions on X ๐คฏ
People want to see templates that:
โ Stop the scroll
โ Persuade the shopper
โ Ultimately maximize RPV
Now, we're democratizing these offers for everyone.
Comment 'DM' and I'll send you the link!
In the last month, @aftersell has had over 1 MILLION impressions on X ๐คฏ
People want to see templates that:
โ Stop the scroll
โ Persuade the shopper
โ Ultimately maximize RPV
Now, we're democratizing these offers for everyone.
Comment 'DM' and I'll send you the link!
I went through data from 50,000+ of brands running
@aftersell post-purchase offers
these 6 templates are CONVERTING harder than everything else.
RT + comment "send it" and I'll DM you the templates
@natelagos Interesting take, Iโve yet to see a brand do it.
My educated guess is that the gap between that first purchase and second purchase is quite large. A lot of people, even with a good product experience, may not put in the effort/cost of reordering to make the tradeoffs worth
This is just the beginning
We are adding Watson to:
- Client Slack channels
- Customer support
- MCP Integration
Shoutout to @tobi and @victorcardenas - River and Twin were our inspirations.
Comment if you want to be on the waitlist ๐ค
An AI agent has been fixing bugs at @aftersell for 6 months.
It costs us $400/mo.
Meet Watson๐
We audited all of our engineering tickets last year and 35% weren't code bugs. They were config questions, duplicates of bugs we already knew about, confusing UI, etc.
And then the real bugs? The fixes are always easy but investigation was always the inefficient part.
Watson now finds and squashes bugs and has started to implement features fully autonomously. Here's how it works:
1) Triage - is this worth looking into?
Watson classifies every ticket before it digs in - is it a config issue, known issue, does it need more info? Output is structured data so downstream stages branch off of a real contract instead of parsing what a model is saying. Helps a lot with latency, cost, and performance.
2) Investigate - what happened?
This is the core of why we built Watson - it'll reconstruct the entire event timeline, read the codebase, pull cloud logging, etc. And we built this all directly on top of the Claude Platform Client SDK so we can have full control over prompt caching, per-tool call caps, cost accounting, etc. It's a super powerful way to control cost - on a "warm run" - 98% of inputs are from the cache.
3) Implement - write to the codebase
Watson directly writes to the repo, makes a PR, etc. It also monitors and reviews PR comments from reviewers. This is the simplest part - writing the code is so simple when you've gotten all of the context from triaging and investigation.
4) Memory - self improving loop (MUST HAVE)
Watson has a persistent memory mounted via GCS Fuse. Every time Watson is invoked it brings back findings and artifacts. We're layering it into episodic (past investigations), semantic (distilled patterns and playbooks), and reflective (what investigation strategies actually work on this codebase).
Watson integrates into everything we have - all internal chats, support systems, Linear, Github, etc. It's constantly improving and every single Linear ticket we have gets looked at by Watson.
Our first run was ~$50 and now it costs us $3.87 per ticket all in.
Comment below if you want me to write up a guide for how to build your own Watson.