Swetha Sree kannan

[Swetha Sree kannan @swethasree93]

So the ask is not fewer people using AI. It is more humans wearing more powerful thinking hats, with a responsibility hat on top. The call itself is the skill now: what goes to the model, and what still needs you.

[Swetha Sree kannan @swethasree93]

Google has been watermarking Gemini's text since 2024. Anthropic committed this week that new Claude models will mark theirs from day one, with an invisible watermark that travels with copy and paste. Under the EU AI Act, every provider serving Europe now has to follow. 🧵

[Swetha Sree kannan @swethasree93]

What earns citations today is original evidence: real numbers, real comparisons, a named author who can be checked. The playbook does not change. Be the source, not the summary.

[Swetha Sree kannan @swethasree93]

Here is my favourite detail in the whole regulation. The EU wrote the exemption itself: disclosure of AI-generated text can be lifted when a human has reviewed it and someone holds editorial responsibility for it. The law is literally asking for a responsibility hat.

[Swetha Sree kannan @swethasree93]

Content: the delivery pipes stay the same. We read hundreds of pieces of content every week, and the ones that 'wow' us still have human thinking driving them. Someone is willing to defend the point of view. No watermark can detect that, and no reader ever misses it.

[Swetha Sree kannan @swethasree93]

Confidential data: this is where the teeth are. Feeding client data into AI tools was always a no-no. The difference now is that obligations reach the teams deploying the tools, not just the companies building them.

[Swetha Sree kannan @swethasree93]

SEO and AEO: it changes less than the panic suggests. Google has watermarked its own model's text for two years and has never confirmed using it in ranking. No answer engine has announced filtering sources by watermark either.

[Swetha Sree kannan @swethasree93]

I read it as governance catching up to the speed of AI adoption. The tools shipped in weeks. The rules took years. This month they finally drew level.

For marketing teams, I see it landing in three places.

[Swetha Sree kannan @swethasree93]

@Bradgohtrades Tutorial

[Swetha Sree kannan @swethasree93]

@FidelCacheFlow buyer

[Swetha Sree kannan @swethasree93]

@pierreeliottlal @romanbuildsaas GOJI

[Pierre-Eliott Lallemant @pierreeliottlal]

GojiberryAI just crossed $300k MRR.

Building GojiberryAI taught us something surprising.

Most AI startups never make it past $1k MRR.

A few scale to millions.

The difference usually isn't better features.

It's two simple things almost everyone underestimates.

If you're building an AI startup, this might save you months of trial and error.

To celebrate, we put together a free course breaking down exactly how we went from $0 to $4M ARR in under a year.

Comment GOJI and repost this post, and I'll DM you the course.

[Swetha Sree kannan @swethasree93]

@rom1trs Edit

[Swetha Sree kannan @swethasree93]

@yanndine GTM

[Swetha Sree kannan @swethasree93]

@borjafat Skill

[Swetha Sree kannan @swethasree93]

@tim_yakubson Stack

[Swetha Sree kannan @swethasree93]

@rom1trs Vox

[Swetha Sree kannan @swethasree93]

@yanndine @AlfieJCarter Skills

[Swetha Sree kannan @swethasree93]

@romanbuildsaas Claude