The World Cup has begun and the group chat is arguing over who's the MVP team and player ⚽️
In AI that one's settled.
The flashy part is the agents that book, buy and negotiate for you...........but an agent is only ever as skilled as the data it trained on.
The data's the MVP.
Always was.
It's Friday 👉 you're logging off.
Your AI agent isn't.
While you're out enjoying life, agents are settling payments, checking balances and getting work done on-chain.
Weekends are for humans... not for agents.
We are ready for @superai_conf day 2🇸🇬
Come and say hi to us! Let's see if we can explore something together in AI.
Or would you rather talk to our agent?
@Zac_labs@superai_conf@MiniMax_AI From what I see, AI products that can serve enterprises and support customer payments are going to be in huge demand over the next 12 months.
Pundi X’s @superai_conf first day takeaway
- Mistral is the big sponsor, but Anthropic and OpenAI visibly absent. @MiniMax_AI cofounder onstage.
- Alibaba is the only major company that has agentic payment wide roll out amongst consumers. Many millions used it to buy milk tea on Qwen app.
- Val from @weka is a semi analysis twin, lots of KV cache goodies in both his session.
- web3 guys were visibly less than last year’s SuperAI. Most in this year’s event are serious ones, I think.
- realisation : corporates are using Microsoft Teams and 365 and have no plans to switch so many corporate will be copilot clients.
@superai_conf Learning directly from the teams at @Zai_org , @minimax, and @MistralAI!
Our team is actively exploring new tools and models, and the next few days at the conference are going to be exciting.
Pundi X team is at @superai_conf this week🇸🇬
Meeting founders, investors, and AI builders, sharing ideas, and exploring what is next for AI.
See you there!
Kicking off the week with this repost about the future of agentic commerce 🌐
@stripe laid out 5 levels of agentic commerce in their annual letter, borrowed from how we talk about self-driving cars.
Each level is defined by what you stop doing, not what the AI starts doing.
1/ You choose the items, the agent handles checkout. Filling forms, not making decisions.
2/ You describe what you need, the agent researches and brings back options. You still make the call.
3/ You delegate a task ("coffee beans under $10") and the agent finds it and buys it, already knowing your preferences.
4/ The agent runs the recurring stuff: restocking, subscriptions, inside the rules you set.
5/ Last one: No prompt at all. The agent anticipates what you need and buys it.
By Stripe's own read, most of us are still at Level 1 and 2.
The jump that matters comes higher up: an agent that arrives with intent and payment authority before a brand ever runs its funnel.
🎥 @luizabr1402, BD @ Pundi AI.
AI models are only as good as the data behind them.
The next frontier isn’t more data… it’s trusted, verified & real-world data that AI can actually rely on.
::: quality > quantity.