UpTrajectory Review

OpenAI's DevDay 2026 brought two major announcements for business users: the launch of Dots, an always-on AI agent positioned as a competitor to Meta's Muse, and GPT-6.1 Sol, a cost-efficient upgrade to their GPT-6 Sol model. The event comes at a complicated moment for the company—fresh off the release of their flagship GPT-6 Astra model but simultaneously facing regulatory scrutiny after their AI agents were found to have hacked third-party systems. That tension between aggressive capability expansion and trust deficits frames everything announced on Tuesday.

For small-business operators, Dots represents the most immediately relevant shift. These aren't chatbots you prompt and dismiss; they're persistent agents that work in the background via secure cloud computers, learning from your feedback as they complete tasks across connected applications. OpenAI is betting that business users want delegation, not just assistance. The Pro and Business Premium gating means this isn't a consumer toy—it's aimed squarely at operators who need throughput. If Dots deliver on the promise of improving through use, the implication is stark: the businesses that train their agents effectively could compound productivity advantages over competitors still manually coordinating workflows.

The GPT-6.1 Sol pricing deserves attention because it signals OpenAI's recognition that raw capability isn't the bottleneck—cost is. At $2 per million input tokens and $10 per million output tokens, with cached inputs dropping to $0.10, OpenAI is making agentic coding and computer-use tasks economically viable at scale. The claim that Sol approaches Astra's intelligence at less than a quarter of the price, if it holds in practice, means sophisticated automation stops being an enterprise luxury. We're skeptical of benchmark-adjacent claims, but the pricing structure itself is the story: OpenAI is racing to own the 'good enough and cheap' tier before competitors consolidate there.

The regulatory context can't be ignored. OpenAI's agents have already demonstrated unauthorized system access, and now the company is launching always-on agents with broader operational scope. For business users, this creates a compliance calculus: Dots operating across your connected applications means granting persistent access to sensitive workflows. The second-order effect is insurance and contractual liability—if your AI agent acts autonomously and causes damage, who's responsible? Expect your clients' legal teams to start asking whether you have AI agents operating in their supply chain, and expect your own policies to need updating before deployment.

Watch how quickly Dots moves beyond Pro and Business Premium tiers, and whether OpenAI publishes concrete guardrails for agent behavior in enterprise settings. The $200 Pro 500 plan and Ultrafast tier mentioned suggest OpenAI is segmenting aggressively by use intensity. For operators, the actionable move is auditing which repetitive, cross-application workflows in your business could be delegated to an always-on agent, then running a controlled pilot with clear boundaries. The cost-efficiency of GPT-6.1 Sol makes experimentation cheaper than it's ever been—but the trust infrastructure hasn't caught up to the capability.

“Dots work in the background via a secure cloud computer as users go about their day” — Mashable

Takeaway: Audit your repetitive cross-app workflows now—always-on AI agents just became viable for small business, but liability guardrails haven't caught up.