The companies that get you to trust an AI with your bank password win the next decade of computing.
The Summary
- OpenAI launched Dots, an "always on" personal AI agent at DevDay, following Meta's Muse and Instinct launches — marking three major agent releases in as many months
- Executives from Anduril, KPMG, and Adobe are already using personal agents for expense reports, hiring dog walkers, booking flu shots, and negotiating insurance deals
- Unlike Meta's free offerings, Dots is paywalled behind Enterprise, Business Premium, and ChatGPT Pro subscriptions — OpenAI also announced a new $500/month top-tier plan
- The common denominator across early adopters: they want AI to do the work, but they still want final approval before action
The Signal
OpenAI's timing is deliberate. After watching Instinct and Meta's Muse grab headlines, the company responded with Dots at its annual DevDay conference in San Francisco. The message is clear: personal agents are the next battleground, and whoever gets you to connect your bank account, email, and calendar first wins your loyalty for the long haul.
This isn't just about productivity tools. It's about lock-in. When an AI knows your insurance provider, your dog walker's schedule, and the six months of expense reports you've been avoiding, switching costs become existential. Jeff Miller, Anduril's CMO, is already "on Muse, ChatGPT, you name it" — using these tools to finally clear expense backlogs and hire help. That's the promise: the AI does what you've been putting off.
"The era of personal agents may finally be upon us — and business leaders are embracing it."
But there's a pattern in how executives are actually using these tools. They're not handing over full autonomy. Every executive interviewed by Business Insider wanted final approval before AI took action. Book the dog walker, sure. But let me click "confirm" first. Negotiate my insurance, great. But show me the offer before you accept it.
This approval-gate tension is the real story. The tech companies want you dependent on agents that act while you sleep. The early adopters want assistants that present options, not make decisions. That gap is where billions in product design decisions will be made over the next 18 months.
Key adoption signals from executives:
- Expense reports and administrative backlog are the killer early use case
- Personal errands (dog walkers, flu shots) are where trust builds
- Financial negotiations (insurance deals) show willingness to give agents real stakes
- But approval workflows remain non-negotiable for most
OpenAI's pricing strategy tells you what they think this is worth. Dots isn't free like Meta's offerings. It's Enterprise, Business Premium, and Pro subscribers only. The new $500/month top-tier plan signals they're betting companies will pay serious money for agents that integrate deeply enough to actually save time at scale.
The announcement also included ChatGPT Space for team collaboration, which matters because personal agents only scale if they can work alongside your coworkers' agents. This is infrastructure for a world where your AI and my AI need to schedule a meeting without either of us touching a keyboard.
The Implication
If you're waiting to see which personal agent wins before jumping in, you're playing it safe and maybe missing the point. The real question isn't which AI is best today. It's which one you'll trust with your credentials first, because that's the one that'll learn your patterns, anticipate your needs, and make switching to a competitor feel like moving cities.
For companies building agents, the mandate is clear: win trust through constrained, high-value tasks first. Expense reports, not mortgage applications. Dog walkers, not divorce lawyers. Build approval workflows that feel like safety rails, not speed bumps. The executives using these tools today are your early majority tomorrow — and they're telling you exactly what they need.