Wednesday · Sep 30, 2026
·
The Signal · Issue No. 59
Notable
Daily AI intelligence, translated for business operators.
Today's Signal
OpenAI's always-on agents are now live in ChatGPT. Your tools just got a 24/7 employee.
OpenAI launched Dots, always-on agents that work around the clock from the cloud, integrated into ChatGPT, Slack, and Teams. They connect to over 4,000 apps and don't count against usage limits. This is the first time a frontier AI model is powering a 24/7 agent category that competitors like Grok Bot and Muse already proved works.
What this means for you
You can now delegate repetitive workflows to an AI that runs even when you do not check in. The question is not whether to use agents but whether you have the operational discipline to define what tasks actually should run unattended in your business.
See how we build it in →
02Models
GPT-6.1 Sol costs a fifth of the top model and closes the gap fast.
OpenAI released GPT-6.1 Sol at $2 per million input tokens and $10 per output tokens, versus Astra's full price. Early benchmarks show Sol performs nearly as well as Astra on many tasks despite the cost cut. This is how frontier AI becomes standard infrastructure for most businesses.
For you: You can now build significant workflows on cheaper models without sacrificing quality, which changes unit economics for any AI-dependent operation.
Go deeper →
03Platforms
ChatGPT Space and Pages give teams a shared AI workspace.
OpenAI added team collaboration hubs where your dots and colleagues work together in a unified space with co-editable docs. ChatGPT now answers inside Slack and Teams threads without leaving those apps. The always-on agent can see and act on conversations in real time.
For you: Your team's context is finally in one place where the AI actually operates, reducing handoffs and the friction of moving information between tools.
Go deeper →
04Tools
The Decisions API gets AI to pick answers in 150 milliseconds.
OpenAI introduced a Decisions API with GPT-6 Luna that selects from preset answers in roughly the time a human blinks. This is OpenAI's version of TypeSafe constraint logic for real-world workflows where you need speed plus safety.
For you: For high-frequency decisions like order routing, customer tier assignment, or content approval, AI can now be fast and reliable enough to run production systems.
Go deeper →
05The Market
AI accelerated marketing speed but not marketing quality, new research shows.
A MarTech analysis found that while AI tools have made marketing execution faster, they have not yet improved the quality or effectiveness of campaigns. Many teams are shipping faster but not thinking harder about what actually moves the needle.
For you: Do not assume faster execution is better execution. Define what success looks like in your marketing before you add an agent to do the work.
Go deeper →
06Signal Check
The model got better. Your operating system did not.
OpenAI shipped GPT-6, Dots, 20+ features, and price cuts in a single day. None of that matters if your business has no way to route tasks to the agent, define what it should do, or measure whether it actually worked. The model is the engine. The operating system is the car. Most businesses are still building the car.
InGen editorial · Sep 30