obsessed with holistic health and optimism
Joined November 2024
- Tweets794
- Following227
- Followers46
- Likes3.8K
OpenAI should just pivot to products. They are getting absolutely destroyed by Anthropic on every level
The Codex usage cut is not nearly as bad as it seems. 6.1 Sol is a very solid model, much, much better than 6 Sol at a very reasonable cost per token
In the end of Summer 2025 everyone was talking about how we are in an AI bubble.
The pace of innovation from that time to now is staggering.
Can’t figure out Opus 5.5. On the surface it seems amazing.
But some of the output quality has been pretty garbage and it feels dumber than fable.
The model is not your moat.
Your data is not your moat either.
Your moat is the system that turns proprietary data into useful intelligence without giving up control of it.
That has been one of the clearest lessons from our work at Eros Innovation.
The strongest AI model today may not be the strongest model six months from now. Long-term AI advantage will not come from committing to one provider. It will come from controlling the data, context and decision system around those models.
Data sovereignty means knowing where your data lives, who can access it, what it can be used for, which rights apply and how its use is recorded.
AI sovereignty means being able to use the best approved intelligence available without surrendering control of your private data, policies, decisions or actions.
This does not mean every organization should build its own large AI model. It means you can adopt better models as they emerge while preserving the parts of the system that belong to you.
One of the biggest misconceptions is that having a large amount of data automatically creates an AI advantage.
It does not.
Data becomes valuable when:
• You know what you have
• You have the right to use it
• It is clean, structured and accessible
• Its relationships and context are preserved
• Sensitive information has clear restrictions
• You can measure whether it improves real outcomes
A million files spread across inboxes, shared drives and internal systems are not an AI moat.
A smaller, rights-cleared and well-governed dataset connected to important business workflows can be.
The practical path is clear:
- Inventory your data.
- Classify it by sensitivity, ownership, rights, location and permitted use.
- Organize the knowledge and relationships inside it
- Put a control layer between your data, users and external AI models.
- Connect multiple approved models instead of depending on one provider.
- Select the right model for each task based on quality, cost, speed, security and data-location requirements.
- Verify the outcome, record what happened and use that evidence to improve the system.
The principle I keep coming back to is simple:
Build what compounds. Rent what commoditizes.
Use the strongest general AI models available. But own the data, cultural or industry context, permissions, workflows, verification and institutional knowledge that make the intelligence uniquely valuable to your organization.
At Eros, this is why we are treating cultural intelligence, rights, policy, model selection, verification and audit as one connected system.
The model provides intelligence.
The organization must still control what that intelligence can access, what it is allowed to do and whether the outcome can be trusted.
That is the real opportunity behind sovereign AI
Using AI requires a lot of taste. For example on the Surface Claude and GPT may feel similar but what they excel in is WILDLY different.
Truly believe you are not maximizing your AI output unless you have at least two subscriptions.
So tired of the cinematic AI product launch videos.
Starting to think throwing up a wiki would be a better differentiator
Idea for Flock. Work with the government to allow everyone within a certain distance to vote on any given potential camera location.
Breaking: Anthropic has extended Fable 5 access through July 12. @claudeai @AnthropicAI
Did you know the best product from @AnthropicAI is not @claudeai ? It is actually their focus music.
