Abhishek Iyer

Indie Saas Builder.

- AI SEO agent
- AI Code diff viewer

prev:
FreeText AI / The Review Index / ex-Google search team

[Abhishek Iyer @distantgradient]

@Marie_Haynes @rustybrick @JoyanneHawkins I wonder if it could be some combination of rank trackers + /goto link resolution tripping up search console

[Abhishek Iyer @distantgradient]

Now, if Anthropic and OAI quickly move to ban this capability in the frontier models, it likely means only one thing:

They've been using these capabilities all along...

[Rikiya Takehi @rikiyatakehi]

We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.

We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: https://t.co/bOWCcFhCVq

[Abhishek Iyer @distantgradient]

@benswerd TIL.

Also true for Illya Sutskever and Demis Hassabis!

[Abhishek Iyer @distantgradient]

"growth hacking" has now become same as "larping"

[Will @willbakerships]

i got fired by @wickedguro this morning

so i'm releasing the entire source code to the public

enjoy 👇
https://t.co/hvRSWp8sf9

[Abhishek Iyer @distantgradient]

@Hobo_Web underrated tweet..

was going to look into factors that govern Google's Gemini reranker scores.. any idea if someone has done something similar?

[Abhishek Iyer @distantgradient]

@matt_gray_ the 4 signature phrases seem to matter for ai answers too..

repeate phrase is likely what gets attached to your name

[Abhishek Iyer @distantgradient]

@asaio87 It's not that people don't care ... I think most do .. I think they get tripped up at "humanizing" in a regular cadence since it's not well defined

A well defined workflow makes creating non-commodity content a breeze...

[Abhishek Iyer @distantgradient]

@CliftonSellers true - and it seems to hold in chatgpt answers too..

a person's post reads like a review, the brand's reads like an ad :)

[Abhishek Iyer @distantgradient]

@akashyadav_dev true. and it usually starts where people ask for what you've already built

search and chatgpt will keep sending traffic long after the launch post dies...

[Abhishek Iyer @distantgradient]

@AIGryffindor people with a list search for a tool like "mailchimp alternative for a small list"

a page answering that will keep sending traffic even when the daily posts stop..

[Abhishek Iyer @distantgradient]

@isuryatk @digitaldutta It's not. I think you're still compromised

check status at
https://t.co/rcpZ7NWeol

[Abhishek Iyer @distantgradient]

@RobbyFrank 1. you mentioned before that set targetting "broad". does that mean turning on advantage+ and not having any audience filtering

2. whats your typical cost per conversion in b2b

3. is the configuration single campaign, multiple ad set (with budget set) and single ad per adset?

[Akanksha - Mobile CRO @AkankshaBarma]

Starting my first SEO experiment:

using https://t.co/jgPZ99oYaJ by @distantgradient to publish a blog post every single day on my new domain

the goal:
- grow organic traffic on autopilot
- build topical authority around app growth, paywalls, and monetization
- attract app founders looking to improve their apps conversion rates

will Google reward the consistency or will I be shouting into the void? 💀

time to find out.

[Abhishek Iyer @distantgradient]

@seo_bint_ejamil https://t.co/Ff5KIyjrFK if you're looking for quality, not scale (disclaimer: I'm the founder)

[Abhishek Iyer @distantgradient]

This.

The responses and your content repo IMO is the biggest differentiator between slop and no slop.

The larger the knowledge base grows, the more your content will diverge from generic Wikipedia-like garbage.

(Also slap on a humanizer so that it does'nt get detected as 100% AI)

[Abhishek Iyer @distantgradient]

@benwills wonder what happens if you build 500 relevant links to a brand nobody talks about..

my guess is chatgpt still doesnt pick it

[Abhishek Iyer @distantgradient]

Rank tracking is causing havoc.

I wonder how many poor souls would be optimising content for rank trackers just because some Claude skill decided to look at GSC data...

[Lily Ray 😏 @lilyraynyc]

Interesting: could this be tied to Google's recent anti-scraping measures?

These are 3 queries in GSC with very unnatural patterns - hundreds of thousands/millions of impressions in 12 months with almost 0 clicks.

Impressions stopped showing up in all 3 cases around Sept 21.

[Abhishek Iyer @distantgradient]

@jakobgreenfeld IMO sitemap is a real alpha here..

a batch of new pages usually tells what they're doing even before any announcement

[Abhishek Iyer @distantgradient]

@asmartbear true - real idea "moat" is the conviction around an idea..

one other thing a clone cant copy is the authority of pages already ranking for your category..

[Abhishek Iyer @distantgradient]

@priteshlakhani Google (and ChatGPT) prefers pages that are old-ish, but updated frequently.

Flip it around - think about it from Google's perspective. What should it trust more, a website that was registered yest, or a seasoned website that keeps its content fresh?

https://t.co/9BKUo5Yxay