Building profitable AI SaaS in public AIPeekaboo.com • AI search visibility SubredditSignals.com • Reddit & X customer leads
- Tweets 17,022
- Following 1,649
- Followers 3,394
- Likes 15,775
Pinned Tweet
We almost hit $50,000 MRR 🤯
A year ago, AI search barely mattered.
Today, companies pay us to understand how ChatGPT, Gemini, Claude, Perplexity, and Google AI talk about their brand.
It's been a lot of late nights, mistakes, and customer conversations, but seeing this category emerge in real time has been incredible.
Grateful to everyone who's trusted us along the way.
Next stop: $55k MRR. 🚀
123
12
449
895,625
AI Visibility Friday #2 👀
SaaS founders, drop your product below.
I’ll check how visible you are in ChatGPT + Gemini and reply with the 1 highest leverage thing I’d do to get recommended more.
Doing these publicly today.
#SaaS
35
1
30
1,064
you can run full report and 2 weeks of data for free here
aipeekaboo.com
1
60
Founders will spend 6 months building a feature nobody asked for
then call distribution luck
Talk to people.
Build what hurts.
Show your work.
Distribution starts way before launch.
20
33
551
Most people are using Claude's reasoning levels wrong.
More reasoning ≠ better results.
On FrontierCode, Opus 5.5 actually drops:
Medium: 54.6%
High: 54.0%
XHigh: 51.4%
Max: 54.4%
The better setup?
Let Opus 5.5 route the work.
Simple task → Low
Most work → Medium
Hard problem → High
Long horizon → XHigh
Frontier problem → Max
You can even have it spin up agents with different reasoning levels for different parts of the same task.
Don't max the reasoning.
Route it.
9
21
630
The biggest advantage in startups isn’t money.
It’s being willing to look stupid longer than everyone else.
Ship the weird idea.
Post to 12 followers.
DM the customer.
Build the thing nobody gets yet.
Most people quit before they have enough evidence to be right.
19
24
484
I grew up poor.
So I thought the goal was to make enough money to finally feel safe.
Turns out there’s no number for that.
You just move the goalpost.
Building a company taught me something money never could:
Security isn’t having enough.
It’s trusting yourself to figure it out when you don’t.
18
2
30
531
Founders love solving problems with code because code cannot reject them.
Customers can.
That is why so many startups are overbuilt and under sold.
21
1
28
408
Dam finally Google Making Moves
Announcing Gemini 4 Argon, our new frontier model.
Argon is built to sustain deep reasoning across complex, long-horizon workflows and delivers frontier performance in complex workflows across real-world software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense.
We’re also expanding the model’s output token limit to an industry-leading 1M tokens.
Argon is currently rolling out to a set of trusted cyber defenders in the Fairwind Program, with broader availability as soon as possible.
9
17
666
GEO is starting to feel a lot like SEO in 2005.
Everyone knows it matters.
Almost nobody knows what actually works.
And the people figuring it out right now are going to have an unfair advantage for years.
40
2
47
1,225
Building a startup is mostly choosing which pain you want.
The pain of uncertainty.
The pain of looking stupid.
The pain of failing in public.
Or the pain of wondering what wouldve happened if you actually tried.
Pick your pain.
17
26
450
DevDay was pretty underwhelming.
GPT 6.1 Sol.
Another banked reset.
Ultrafast mode.
And a new $500 plan.
Meanwhile Pro is going from 20x usage to 10x.
Dots are cool I guess, but I wanted better models and more usage, not a more expensive tier.
After all the hype, this feels like another L.
What did you guys think?
16
30
1,118
The best founders I know are slightly unreasonable
They ship before they’re ready
They obsess over things nobody else notices
They keep going long after the excitement wears off
Being reasonable is great for maintaining things
It’s rarely how you build something new
38
45
809
This is just wild.
Sonnet 5.5 is basically knocking on Opus 5.5’s door across the board and actually beats it on Terminal Bench.
The gap between fast model and frontier model is disappearing insanely fast
8
18
504
Your old Claude prompts might be making Opus 5.5 slower
Anthropic tested removing “think carefully” from a chat system prompt
Replies started sooner. No clear drop in quality.
Some prompt engineering is just technical debt now
14
18
435
most feature lists dont matter as much as how fast you ship learn and earn trust with users
thats the real moat not another checkbox on a landing page
17
36
673
