Kathmandu: AI’s Next Phase became clearer this week as OpenAI, Pony.ai, Tesco, and Cursor each introduced products and updates that push artificial intelligence deeper into real-world work. From paid coding tools and self-improving robotaxi systems to retail assistants and multi-agent software development, these announcements show that AI’s Next Phase is less about simple chat and more about systems built for tasks, workflows, and scale.
OpenAI Shows How AI’s Next Phase Is Becoming More Tiered
OpenAI’s role in AI’s Next Phase came into focus with the launch of a new $100 per month Pro plan. The company placed it between Plus at $20 and its existing Pro tier at $200, giving users a middle option designed for more serious work, especially longer Codex sessions.
OpenAI says the new plan offers 5 times higher usage limits than Plus. For a limited time, it also includes up to 10 times more Codex usage than Plus, with that temporary boost lasting until May 31, 2026. In AI’s Next Phase, pricing is no longer just about access. It is increasingly about how much real work a user can do with an AI tool.
Pony.ai Pushes AI’s Next Phase Toward Self-Improving Driving Systems
Pony.ai’s announcement of PonyWorld 2.0 shows another side of AI’s Next Phase. The company says its upgraded world model can identify its own weaknesses, direct targeted data collection, and improve training by focusing on the hardest real-world driving cases.
According to Pony.ai, the system reviews decisions at scale, compares intent with outcomes, and flags where more learning is needed. It then creates data-collection tasks for human teams, who gather real-world examples and send them back into the system. In AI’s Next Phase, AI is not just responding to prompts. It is being built to spot its own limits and help drive its own improvement.
Tesco Brings AI’s Next Phase Into Everyday Shopping
Tesco is also testing what AI’s Next Phase looks like in retail. The company has started a large internal beta of an AI assistant in its app, first giving access to about 280,000 employees before a broader customer launch later in 2026.
The assistant focuses on meal planning, recipe suggestions, and building a shopping basket once a meal is chosen. It can also use shopping history and preferences to make recommendations. For Tesco, AI’s Next Phase is about making grocery shopping easier, more personalized, and potentially less wasteful.
Cursor Turns AI’s Next Phase Into a Multi-Agent Workspace
Cursor’s release of Cursor 3 highlights how AI’s Next Phase is changing software development. The company says the update introduces an agent-first workspace, where users can run many coding agents in parallel and manage handoffs between local and cloud execution.
Cursor also introduced Glass, its new interface for working with agents. The company says it helps users review edits, manage staging and commits, and track work across different repositories. In AI’s Next Phase, developers are moving beyond one assistant in one chat window and into coordinated systems built for real production workflows.
Why AI’s Next Phase Matters
What ties these stories together is simple. AI’s Next Phase is moving away from one-off chatbot experiences and toward products built for heavy use, feedback loops, testing, and coordination. OpenAI is refining usage tiers, Pony.ai is building self-diagnosing systems, Tesco is testing AI at workforce scale, and Cursor is redesigning coding around multiple agents.
AI’s Next Phase is no longer a future idea. It is already showing up in pricing, retail apps, autonomous driving, and software development.
Leave a Reply