OpenAI launched always-on Dots agents and GPT-6.1 Sol at DevDay 2026. Here is what every plumber, dentist, HVAC owner, and contractor needs to know today.
Ido Cohen · Published 2026-09-29 · AI for Service Business
OpenAI just shipped the most consequential product update for small-business owners since ChatGPT launched: at DevDay 2026 in San Francisco today, the company unveiled Dots, always-on autonomous AI agents that keep working after you close the browser window, alongside a new model, GPT-6.1 Sol, that delivers near-flagship performance at one-fifth the price. If you run a plumbing company, dental practice, law firm, HVAC business, med spa, or any other service operation, read this before your competitors do.
Two announcements lead everything else out of OpenAI's more than 20 DevDay reveals.
Dots are always-on AI agents, each with their own cloud computer and browser, powered by GPT-6 Astra — OpenAI's most capable model. According to CNBC's live coverage, OpenAI CEO Sam Altman described them as "remarkably capable, always-on agents built to handle really anything you can think of." Unlike a standard ChatGPT conversation that pauses until you type again, Dots run continuously in the background. They connect to more than 4,000 apps through OpenAI's plugin ecosystem and can be reached via ChatGPT, Slack, and Teams — with texting support coming soon.
GPT-6.1 Sol is a major upgrade to GPT-6 Sol, released just one week ago. TechCrunch reported that the new model delivers "nearly the same level of intelligence as GPT-6 Astra for agentic coding, computer use, and professional work, at one-fifth the standard input and output token prices." That is $2 per million input tokens and $10 per million output tokens, versus Astra's standard pricing. Factual error rates dropped from 11.4% on GPT-6 Sol to 7.7% on the new model at low reasoning effort.
Both are available today to ChatGPT Pro and Business Premium users.
Most AI tools for service businesses are reactive. You open a tab, ask a question, close the tab. Dots flips that model entirely.
According to VentureBeat's DevDay coverage, Dots are "designed to keep working after an employee closes the chat window — monitoring projects, using software, responding to changing information and bringing completed work back for approval." Each Dot has its own cloud computer, its own credentials, and access to the apps you connect. OpenAI's internal use cases shared at the event included a Dot that noticed an invoice had not been sent and drafted it for human approval — without being asked.
Here is what that looks like in a service-business context:
OpenAI noted that "over time, we envision teams of dots working together on your behalf" — meaning you could eventually run a booking Dot, a follow-up Dot, and a review-management Dot simultaneously.
Benchmarks are fine. Price-to-performance ratios are what actually change behavior.
GPT-6.1 Sol costs $2 per million input tokens and $10 per million output tokens according to reporting from The Next Web and SQ Magazine. That is one-fifth of GPT-6 Astra's standard pricing. For context, GPT-6 Sol and Luna launched just one week before DevDay, and OpenAI cut the price again — a sign the AI price war is still accelerating.
What does that mean for a service business using AI-powered tools? Practically, the per-query cost of running automations — lead scoring, appointment confirmations, content drafts, intake summaries — keeps falling. Any AI tool your software vendor builds on top of OpenAI's API will get cheaper to operate. If you are using a CRM with AI-written follow-up sequences, an AI phone answering product, or an automated review-response tool, expect providers to pass some of that cost reduction through as expanded usage tiers or lower subscription prices over the next 90 days.
The model is available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users. It is not yet in the main ChatGPT chat interface.
On the day before DevDay, OpenAI confirmed it decided not to release GPT-6.1 Astra after determining the model "did not adequately meet the company's safety standards," according to CNBC's live updates. Sam Altman characterized this as "the normal course category." That decision is worth noting because it runs counter to the typical pattern in tech of shipping first and patching later.
For service businesses, the safety bar on AI agents matters more than it does for casual chatbot users. A Dot that books the wrong appointment, sends an invoice to the wrong client, or misreads a compliance deadline is not just a nuisance — it is a liability. OpenAI pulling a flagship model for safety reasons, at the cost of positive DevDay headlines, is a signal the company is making at least some grown-up decisions about what it ships.
That said, Dots are still new. Treat the first 30–60 days as a supervised pilot, not a fully autonomous deployment.
The always-on agent category is getting crowded fast. Here is where Dots sits relative to what service businesses might already be evaluating:
The key differentiator for Dots is the persistent cloud computer with a browser — meaning a Dot can navigate web-based tools that don't have an API, not just platforms with pre-built connectors. That is significant for service businesses running older software that has never been "integrated" with anything.
DevDay also introduced ChatGPT Space, a shared workspace where human team members, ChatGPT, and Dots can work from the same project context. VentureBeat described it as a model where "employees and multiple AI agents can work against the same project context, hand work back and forth, and update common deliverables."
For a five-person HVAC office, that is meaningful. Instead of one owner's personal ChatGPT subscription doing everything in isolation, a Space gives the whole team — and their Dots — a shared board. The dispatcher's Dot, the billing coordinator's Dot, and the owner's strategic Dot can all read from the same job list and customer notes without anyone manually copying context between them.
This is early-stage, and the workflow discipline required to set it up correctly is real. But the direction is unmistakable: AI is moving from individual productivity tool to team infrastructure.
You don't need to rebuild your entire operation before Friday. But you do need to start positioning now, because the service businesses that build the first Dot-powered workflows this quarter will have a structural efficiency edge by Q1 2027.
Here is a concrete 5-step sequence:
1. Audit your most repetitive, schedule-driven tasks today. Write down every workflow that runs the same way every time: review requests, appointment reminders, invoice follow-ups, lead nurture sequences, renewal outreach. These are your Dot candidates.
2. Check your ChatGPT plan. Dots are live now for Pro ($20/month) and Business Premium users. If you or your team are on a Free or Plus plan, the upgrade is the prerequisite. Log in and check your current tier at chat.openai.com.
3. Map your app stack against OpenAI's 4,000+ integrations. The Dots value scales with the number of apps it can touch. Before configuring anything, list your CRM, scheduling software, email platform, invoicing tool, and review platform. Confirm which ones have ChatGPT plugins or are in the Zapier/Make ecosystem that connects to OpenAI.
4. Run one supervised pilot in the next 14 days. Choose your lowest-risk workflow — something like a review-request follow-up or a cold-lead re-engagement sequence. Set up a Dot to draft the outreach, but keep human approval required before anything sends. Log the errors and the time saved.
5. Watch GPT-6.1 Sol's rollout into your existing tools. If you use any AI-powered CRM, phone answer service, or marketing tool, check their changelog in the next 30 days. Vendors building on OpenAI's API will likely push GPT-6.1 Sol updates that improve accuracy and reduce per-request cost. You don't have to do anything — just know what's improving under the hood.
---
What is OpenAI Dots and how is it different from regular ChatGPT?
Dots are persistent AI agents that run continuously in the background on their own cloud computer, even when you are not actively chatting. Regular ChatGPT stops working the moment you close the conversation. Dots keep pursuing assigned goals, monitor connected apps, and bring finished work back to you for review — 24/7.
Do I need a developer to set up a Dot for my service business?
Not necessarily. OpenAI designed Dots for non-technical users, with setup flowing through ChatGPT's existing interface. However, connecting a Dot to a specific CRM or industry-specific tool may require someone comfortable with the app's integration settings. Start with broadly supported apps like Google Workspace, Slack, or Salesforce before attempting niche software.
Is GPT-6.1 Sol available in the free version of ChatGPT?
No. As of today, GPT-6.1 Sol is available to Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex. It is not yet available in the standard ChatGPT chat interface or on the Free tier. OpenAI has not announced a timeline for broader rollout.
Should I be concerned about a Dot sending a wrong invoice or booking the wrong client?
Yes, and that concern is valid. Always-on agents that can take real-world actions in your business systems carry real liability risk if misconfigured. For your first 30–60 days, run any Dot in approval-required mode: it drafts and queues, but nothing sends until a human reviews it. Treat Dots like a new hire who needs supervision before working unsupervised.
Why did OpenAI cancel GPT-6.1 Astra one day before DevDay?
According to CNBC's DevDay live updates, OpenAI determined the model did not meet its internal safety standards and chose not to release it. Sam Altman characterized this as routine rather than exceptional. The decision signals that at least some guardrails exist on the release pipeline — which matters more for autonomous-agent products like Dots than for standard chat models.
---
Sources: