Claude for writing. GPT for structure.
Gemini for what happened today. Grok for the take nobody asked for.
You don't have a favorite child.
You shouldn't have a favorite model either.
๐ https://x.com/AnumaAI/status/2089427022019748199
Gemini for what happened today. Grok for the take nobody asked for.
You don't have a favorite child.
You shouldn't have a favorite model either.
๐ https://x.com/AnumaAI/status/2089427022019748199
X (formerly Twitter)
Anuma (@AnumaAI) on X
Claude for writing. GPT for structure.
Gemini for what happened today. Grok for the take nobody asked for.
You don't have a favorite child.
You shouldn't have a favorite model either.
Gemini for what happened today. Grok for the take nobody asked for.
You don't have a favorite child.
You shouldn't have a favorite model either.
"Deleted" should mean deleted.
Not "hidden from you but kept for us."
Wild that this is a differentiator in 2026, but here we are. #PrivateAI
๐ https://x.com/AnumaAI/status/2089774306096005351
Not "hidden from you but kept for us."
Wild that this is a differentiator in 2026, but here we are. #PrivateAI
๐ https://x.com/AnumaAI/status/2089774306096005351
X (formerly Twitter)
Anuma (@AnumaAI) on X
"Deleted" should mean deleted.
Not "hidden from you but kept for us."
Wild that this is a differentiator in 2026, but here we are. #PrivateAI
Not "hidden from you but kept for us."
Wild that this is a differentiator in 2026, but here we are. #PrivateAI
Privacy policies are promises.
On-device encryption is architecture.
One can change in a quarterly update.
The other can't.
Use Claude Privately.
Use ChatGPT Privately.
Use Gemini Privately.
Use Grok Privately. #PrivateAI
๐ https://x.com/AnumaAI/status/2089834701292810601
On-device encryption is architecture.
One can change in a quarterly update.
The other can't.
Use Claude Privately.
Use ChatGPT Privately.
Use Gemini Privately.
Use Grok Privately. #PrivateAI
๐ https://x.com/AnumaAI/status/2089834701292810601
X (formerly Twitter)
Anuma (@AnumaAI) on X
Privacy policies are promises.
On-device encryption is architecture.
One can change in a quarterly update.
The other can't.
Use Claude Privately.
Use ChatGPT Privately.
Use Gemini Privately.
Use Grok Privately. #PrivateAI
On-device encryption is architecture.
One can change in a quarterly update.
The other can't.
Use Claude Privately.
Use ChatGPT Privately.
Use Gemini Privately.
Use Grok Privately. #PrivateAI
cursor is going to space and half of SF is having
a feelings day about their editor
this is why you don't marry a tool
date them all. keep your context. stay flexible.
๐ https://x.com/AnumaAI/status/2089955508476133566
a feelings day about their editor
this is why you don't marry a tool
date them all. keep your context. stay flexible.
๐ https://x.com/AnumaAI/status/2089955508476133566
X (formerly Twitter)
Anuma (@AnumaAI) on X
cursor is going to space and half of SF is having
a feelings day about their editor
this is why you don't marry a tool
date them all. keep your context. stay flexible.
a feelings day about their editor
this is why you don't marry a tool
date them all. keep your context. stay flexible.
๐ฅ1
Fun game: ask the same question in Council Mode and watch four frontier models disagree about something you thought had one answer.
๐พ๐๐๐ฉ ๐๐๐ ๐ซ๐จ. ๐๐ง๐ค๐ ๐ซ๐จ. ๐พ๐ก๐๐ช๐๐ ๐ซ๐จ. ๐๐๐ข๐๐ฃ๐
Confidence is a UI element.
Consensus is information.
๐ https://x.com/AnumaAI/status/2090091767446086102
๐พ๐๐๐ฉ ๐๐๐ ๐ซ๐จ. ๐๐ง๐ค๐ ๐ซ๐จ. ๐พ๐ก๐๐ช๐๐ ๐ซ๐จ. ๐๐๐ข๐๐ฃ๐
Confidence is a UI element.
Consensus is information.
๐ https://x.com/AnumaAI/status/2090091767446086102
X (formerly Twitter)
Anuma (@AnumaAI) on X
Fun game: ask the same question in Council Mode and watch four frontier models disagree about something you thought had one answer.
๐พ๐๐๐ฉ ๐๐๐ ๐ซ๐จ. ๐๐ง๐ค๐ ๐ซ๐จ. ๐พ๐ก๐๐ช๐๐ ๐ซ๐จ. ๐๐๐ข๐๐ฃ๐
Confidence is a UI element.
Consensus is information.
๐พ๐๐๐ฉ ๐๐๐ ๐ซ๐จ. ๐๐ง๐ค๐ ๐ซ๐จ. ๐พ๐ก๐๐ช๐๐ ๐ซ๐จ. ๐๐๐ข๐๐ฃ๐
Confidence is a UI element.
Consensus is information.
โค1
Tech twitter this week: "5.6 Sol is half the price of Fable for the same task"
Correct move:
use Sol for the cheap tasks,
Fable for the hard ones.
One tab > two subscriptions @AnumaAI
๐ https://x.com/AnumaAI/status/2090141980248453473
Correct move:
use Sol for the cheap tasks,
Fable for the hard ones.
One tab > two subscriptions @AnumaAI
๐ https://x.com/AnumaAI/status/2090141980248453473
X (formerly Twitter)
Anuma (@AnumaAI) on X
Tech twitter this week: "5.6 Sol is half the price of Fable for the same task"
Correct move:
use Sol for the cheap tasks,
Fable for the hard ones.
One tab > two subscriptions @AnumaAI
Correct move:
use Sol for the cheap tasks,
Fable for the hard ones.
One tab > two subscriptions @AnumaAI
The new SF status symbol isn't using the best model.
It's knowing which model to use for which task โ
and switching without ceremony.
Model fluency > model loyalty. Anuma is where you
practice it.
๐ https://x.com/AnumaAI/status/2090202126395843068
It's knowing which model to use for which task โ
and switching without ceremony.
Model fluency > model loyalty. Anuma is where you
practice it.
๐ https://x.com/AnumaAI/status/2090202126395843068
X (formerly Twitter)
Anuma (@AnumaAI) on X
The new SF status symbol isn't using the best model.
It's knowing which model to use for which task โ
and switching without ceremony.
Model fluency > model loyalty. Anuma is where you
practice it.
It's knowing which model to use for which task โ
and switching without ceremony.
Model fluency > model loyalty. Anuma is where you
practice it.
"You can run a frontier-ish model on a single 5090 now"
Open models catching up is the best thing happening in AI.
Closed for the edge. Open for privacy and cost.
Both in one app. That was always the endgame.
๐ https://x.com/AnumaAI/status/2090332993659261309
Open models catching up is the best thing happening in AI.
Closed for the edge. Open for privacy and cost.
Both in one app. That was always the endgame.
๐ https://x.com/AnumaAI/status/2090332993659261309
X (formerly Twitter)
Anuma (@AnumaAI) on X
"You can run a frontier-ish model on a single 5090 now"
Open models catching up is the best thing happening in AI.
Closed for the edge. Open for privacy and cost.
Both in one app. That was always the endgame.
Open models catching up is the best thing happening in AI.
Closed for the edge. Open for privacy and cost.
Both in one app. That was always the endgame.
Your AI conversations are either
an asset you own
or a dataset someone else does.
There's no third option. Pick.
๐ https://x.com/AnumaAI/status/2090469386595312106
an asset you own
or a dataset someone else does.
There's no third option. Pick.
๐ https://x.com/AnumaAI/status/2090469386595312106
X (formerly Twitter)
Anuma (@AnumaAI) on X
Your AI conversations are either
an asset you own
or a dataset someone else does.
There's no third option. Pick.
an asset you own
or a dataset someone else does.
There's no third option. Pick.
The re-explaining economy is massive.
Millions of people, every morning,
typing their context into a chatbot that met them yesterday.
We simply stopped participating.
๐ https://x.com/AnumaAI/status/2090564251719029045
Millions of people, every morning,
typing their context into a chatbot that met them yesterday.
We simply stopped participating.
๐ https://x.com/AnumaAI/status/2090564251719029045
X (formerly Twitter)
Anuma (@AnumaAI) on X
The re-explaining economy is massive.
Millions of people, every morning,
typing their context into a chatbot that met them yesterday.
We simply stopped participating.
Millions of people, every morning,
typing their context into a chatbot that met them yesterday.
We simply stopped participating.
Unpopular opinion: model releases are the least
interesting part of AI now.
The interesting part is that your context can finally
outlive all of them.
๐ https://x.com/AnumaAI/status/2090634979625627820
interesting part of AI now.
The interesting part is that your context can finally
outlive all of them.
๐ https://x.com/AnumaAI/status/2090634979625627820
X (formerly Twitter)
Anuma (@AnumaAI) on X
Unpopular opinion: model releases are the least
interesting part of AI now.
The interesting part is that your context can finally
outlive all of them.
interesting part of AI now.
The interesting part is that your context can finally
outlive all of them.
SF Startups guard their cap table, their code, their roadmap,
then paste all three into a chatbot with a free tier.
๐ข๐ป-๐ฑ๐ฒ๐๐ถ๐ฐ๐ฒ ๐ผ๐ฟ ๐ถ๐ ๐ฑ๐ถ๐ฑ๐ป'๐ ๐๐๐ฎ๐ ๐๐ผ๐๐ฟ๐. #PrivateAI
๐ https://x.com/AnumaAI/status/2090785981699956887
then paste all three into a chatbot with a free tier.
๐ข๐ป-๐ฑ๐ฒ๐๐ถ๐ฐ๐ฒ ๐ผ๐ฟ ๐ถ๐ ๐ฑ๐ถ๐ฑ๐ป'๐ ๐๐๐ฎ๐ ๐๐ผ๐๐ฟ๐. #PrivateAI
๐ https://x.com/AnumaAI/status/2090785981699956887
X (formerly Twitter)
Anuma (@AnumaAI) on X
SF Startups guard their cap table, their code, their roadmap,
then paste all three into a chatbot with a free tier.
๐ข๐ป-๐ฑ๐ฒ๐๐ถ๐ฐ๐ฒ ๐ผ๐ฟ ๐ถ๐ ๐ฑ๐ถ๐ฑ๐ป'๐ ๐๐๐ฎ๐ ๐๐ผ๐๐ฟ๐. #PrivateAI
then paste all three into a chatbot with a free tier.
๐ข๐ป-๐ฑ๐ฒ๐๐ถ๐ฐ๐ฒ ๐ผ๐ฟ ๐ถ๐ ๐ฑ๐ถ๐ฑ๐ป'๐ ๐๐๐ฎ๐ ๐๐ผ๐๐ฟ๐. #PrivateAI
Everyone's building an AI moat.
Ours is boring: your data physically isn't on our servers.
Hard to breach what we don't have.
๐ https://x.com/AnumaAI/status/2090861481407443403
Ours is boring: your data physically isn't on our servers.
Hard to breach what we don't have.
๐ https://x.com/AnumaAI/status/2090861481407443403
X (formerly Twitter)
Anuma (@AnumaAI) on X
Everyone's building an AI moat.
Ours is boring: your data physically isn't on our servers.
Hard to breach what we don't have.
Ours is boring: your data physically isn't on our servers.
Hard to breach what we don't have.
AI labs are filing S-1s now.
Reminder: every model you love answers to a board, a roadmap, and a quarterly earnings call.
Your context shouldn't.
Keep your memory hidden.
๐ https://x.com/AnumaAI/status/2090921865191559385
Reminder: every model you love answers to a board, a roadmap, and a quarterly earnings call.
Your context shouldn't.
Keep your memory hidden.
๐ https://x.com/AnumaAI/status/2090921865191559385
X (formerly Twitter)
Anuma (@AnumaAI) on X
AI labs are filing S-1s now.
Reminder: every model you love answers to a board, a roadmap, and a quarterly earnings call.
Your context shouldn't.
Keep your memory hidden.
Reminder: every model you love answers to a board, a roadmap, and a quarterly earnings call.
Your context shouldn't.
Keep your memory hidden.
Know Your Agent (KYA) will become foundational infrastructure.
An agent needs more than an identity. It needs persistent context, explicit permissions, and an auditable history of what it has done.
That's where private, user-owned memory becomes critical. #PrivateMemory #AIMemory
๐ https://x.com/AnumaAI/status/2091136540256952519
An agent needs more than an identity. It needs persistent context, explicit permissions, and an auditable history of what it has done.
That's where private, user-owned memory becomes critical. #PrivateMemory #AIMemory
๐ https://x.com/AnumaAI/status/2091136540256952519
X (formerly Twitter)
Anuma (@AnumaAI) on X
Know Your Agent (KYA) will become foundational infrastructure.
An agent needs more than an identity. It needs persistent context, explicit permissions, and an auditable history of what it has done
An agent needs more than an identity. It needs persistent context, explicit permissions, and an auditable history of what it has done
New model every three weeks. New "GPT killer" every month.
Exhausting if you're loyal.
Free entertainment if you're not.
Stay agnostic. Keep your memory. Let them fight.
๐ https://x.com/AnumaAI/status/2091215555646673192
Exhausting if you're loyal.
Free entertainment if you're not.
Stay agnostic. Keep your memory. Let them fight.
๐ https://x.com/AnumaAI/status/2091215555646673192
X (formerly Twitter)
Anuma (@AnumaAI) on X
New model every three weeks. New "GPT killer" every month.
Exhausting if you're loyal.
Free entertainment if you're not.
Stay agnostic. Keep your memory. Let them fight.
Exhausting if you're loyal.
Free entertainment if you're not.
Stay agnostic. Keep your memory. Let them fight.
Four models. One question.
One summary where they agree and split.
#CouncilMode is basically hiring a panel of experts who bill $0/hour and never do a kickoff call.
๐ง๐ฅ๐ฌ ๐๐ง. ๐
๐ https://x.com/AnumaAI/status/2091586248104145346
One summary where they agree and split.
#CouncilMode is basically hiring a panel of experts who bill $0/hour and never do a kickoff call.
๐ง๐ฅ๐ฌ ๐๐ง. ๐
๐ https://x.com/AnumaAI/status/2091586248104145346
X (formerly Twitter)
Anuma (@AnumaAI) on X
Four models. One question.
One summary where they agree and split.
#CouncilMode is basically hiring a panel of experts who bill $0/hour and never do a kickoff call.
๐ง๐ฅ๐ฌ ๐๐ง. ๐ https://t.co/WNkM8wlTjf
One summary where they agree and split.
#CouncilMode is basically hiring a panel of experts who bill $0/hour and never do a kickoff call.
๐ง๐ฅ๐ฌ ๐๐ง. ๐ https://t.co/WNkM8wlTjf
Instinctโs powerful AI assistant is raising privacy and security concerns.
A tester disconnected the AI from her email at 11am.
At 2pm it summarized her inbox anyway. The emails were already stored in plain text.
"Disconnect" should mean gone.
๐ https://x.com/AnumaAI/status/2092085674165854231
A tester disconnected the AI from her email at 11am.
At 2pm it summarized her inbox anyway. The emails were already stored in plain text.
"Disconnect" should mean gone.
๐ https://x.com/AnumaAI/status/2092085674165854231
X (formerly Twitter)
Anuma (@AnumaAI) on X
Instinctโs powerful AI assistant is raising privacy and security concerns.
A tester disconnected the AI from her email at 11am.
At 2pm it summarized her inbox anyway. The emails were already storโฆ
A tester disconnected the AI from her email at 11am.
At 2pm it summarized her inbox anyway. The emails were already storโฆ
There is no single โbestโ AI image generator.
Midjourney for cinematic visuals.
Ideogram for text.
FLUX for photorealism + control.
Recraft for design.
GPT Image for conversational creation.
We tested 10 of them so you don't have to.๐
๐ https://x.com/AnumaAI/status/2092443621052031435
Midjourney for cinematic visuals.
Ideogram for text.
FLUX for photorealism + control.
Recraft for design.
GPT Image for conversational creation.
We tested 10 of them so you don't have to.๐
๐ https://x.com/AnumaAI/status/2092443621052031435
X (formerly Twitter)
Anuma (@AnumaAI) on X
There is no single โbestโ AI image generator.
Midjourney for cinematic visuals.
Ideogram for text.
FLUX for photorealism + control.
Recraft for design.
GPT Image for conversational creation.
We โฆ
Midjourney for cinematic visuals.
Ideogram for text.
FLUX for photorealism + control.
Recraft for design.
GPT Image for conversational creation.
We โฆ
Everyone has a favorite AI model. Until they try another one. ๐
We tracked 4,471 model switches across real Anuma users. The biggest net losers werenโt necessarily โbadโ models.
They were justโฆ not the model people wanted for that task.
Maybe the future of AI isn't picking a winner. Maybe it's knowing when to switch.
๐ https://x.com/AnumaAI/status/2092628610444128425
We tracked 4,471 model switches across real Anuma users. The biggest net losers werenโt necessarily โbadโ models.
They were justโฆ not the model people wanted for that task.
Maybe the future of AI isn't picking a winner. Maybe it's knowing when to switch.
๐ https://x.com/AnumaAI/status/2092628610444128425
X (formerly Twitter)
Anuma (@AnumaAI) on X
Everyone has a favorite AI model. Until they try another one. ๐
We tracked 4,471 model switches across real Anuma users. The biggest net losers werenโt necessarily โbadโ models.
They were justโฆโฆ
We tracked 4,471 model switches across real Anuma users. The biggest net losers werenโt necessarily โbadโ models.
They were justโฆโฆ
13 major AI platforms. One uncomfortable takeaway: The AI you trust with your most personal thoughts may be collecting, sharing, or using more data than you realize.
Privacy isnโt a feature you should have to hunt for in a 40-page policy.
The best and worst AI for your privacy, ranked, and how each handles your data:
๐ https://x.com/AnumaAI/status/2092820847207571939
Privacy isnโt a feature you should have to hunt for in a 40-page policy.
The best and worst AI for your privacy, ranked, and how each handles your data:
๐ https://x.com/AnumaAI/status/2092820847207571939
X (formerly Twitter)
Anuma (@AnumaAI) on X
13 major AI platforms. One uncomfortable takeaway: The AI you trust with your most personal thoughts may be collecting, sharing, or using more data than you realize.
Privacy isnโt a feature you sโฆ
Privacy isnโt a feature you sโฆ