Selali Adobor

building a consumer product

[Selali Adobor @notionsmith]

nano-banana is a reasoning model!

...kind of.

The only difference between these two outputs is prompting the model to think in the right way: (🧵)

[Selali Adobor @notionsmith]

@benln building consumer 👋🏾

[Selali Adobor @notionsmith]

I noticed the wombat hops when you click it and had a thought.
https://t.co/6sgfZsvBGm

[Selali Adobor @notionsmith]

all in good stride though! just a reminder to stay focused on building something people want. 🙂

[Selali Adobor @notionsmith]

@Boudatw that sounds a lot like https://t.co/vgHb83XwIg

[Selali Adobor @notionsmith]

@adrielyong @a16z @speedrun Just sent a DM, building community-driven consumer AI and would love to talk!

[Selali Adobor @notionsmith]

@brianyunmd @zebird0 I'm working on Spellbound to help build AI that tells great stories, I just sent a DM!

[Selali Adobor @notionsmith]

@aryanvs_ @a16z I'm building consumer AI infra, would love to talk in both business and ML terms!

[Selali Adobor @notionsmith]

@Boudatw

[Selali Adobor @notionsmith]

@thechangj building infra for consumer AI: would love to talk!

[Selali Adobor @notionsmith]

For nano-banana, steering towards structured thinking works better than an open-ended request to think:

"You must think at extremely great length, for several hundred words, focusing on having a structured and thorough thinking process, before returning the image. Think in the 1st person, and in terms of what the user has asked, then return the generated image."

The length adds some latency and cost here, but can make a big difference for spatial understanding ⬇️

[Selali Adobor @notionsmith]

nano-banana is a reasoning model!

...kind of.

The only difference between these two outputs is prompting the model to think in the right way: (🧵)

[Selali Adobor @notionsmith]

Hybrid models are the new default, so recent non-reasoning models often share some post-training with their reasoning counterparts.

So they come up with higher quality reasoning traces than past models:

[Selali Adobor @notionsmith]

@Boudatw ... and that's why I'm building Spellbound: to create ai that tells great stories. dm in your inbox 😉

[Selali Adobor @notionsmith]

@Boudatw @speedrun sent a message, would love to know if I need to brag more 😂

[Selali Adobor @notionsmith]

@brycent Sent a DM!

[Selali Adobor @notionsmith]

@sama Is this Horizon Alpha?

[Selali Adobor @notionsmith]

@MarcelFromMimic @ycombinator @garrytan programmer trying to learn more about design: is there a typography reason the text lines up like this?

(and how did you decide on indenting the subtitle)

[Selali Adobor @notionsmith]

@michellechen Is there any reason why Llama 4 was a priority over Gemma 3? Gemma 3 seems like a much better fit to the Workers AI architecture and value prop

(updated Gemma w/ LORA support would also add a lot more value to the feature than the current base models supported)

[Selali Adobor @notionsmith]

(more seriously, I believe the big winner in AI Wattpad won't be built on agentic flows. If you focus on one-shot performance you get content that isn't as shareable, but feels much more engaging than possible otherwise)

Spellbound's average sessions for AI generated fanfic are exceeding 1 hour because we've focused on post-training for 1 shot performance that gets the reader to turn the next page)

[Selali Adobor @notionsmith]

@Tocelot @speedrun Sounds like you're looking for something a bit more in depth than the stories we craft with https://t.co/vGC1zLkYdO 😔