OpenAI Dots for Sales Teams: Pricing, Limits & First Workflows
OpenAI's DevDay 2026 (September 29) packed more than 20 announcements, but one matters far more than the rest for revenue teams: Dots — always-on ChatGPT agents that run on their own cloud computers, connect to 4,000+ apps, and keep working after you close the tab.
If the last two years were about chatting with AI, Dots are OpenAI's bet that the next two are about delegating to it. Here's what actually shipped, what it costs, where it breaks for sales use cases, and five workflows worth testing this month.
What is a Dot?
A Dot is a persistent agent inside ChatGPT. Unlike a normal chat session, each Dot:
- Runs on its own cloud computer and browser — it can open websites, fill forms, and use web apps without your laptop being on
- Is powered by GPT-6 Astra, the frontier model OpenAI introduced on September 3
- Connects to 4,000+ apps via OpenAI's connector ecosystem
- Works across surfaces — you can hand it tasks from ChatGPT, text message, email, or Slack
- Handles recurring work — it follows up on open tasks and can delegate pieces to sub-agents
The mental model: less "smart autocomplete," more "contractor with a laptop who never logs off."
It wasn't a flawless launch — one Dot famously had a "slow morning" and failed to respond during the live demo. Always-on doesn't mean always-reliable yet, which matters if you're thinking about putting one in a revenue-critical path.
Dots pricing: what it actually costs
| Plan | Price | Dots included |
|---|---|---|
| ChatGPT Free / Go / Plus | $0–$20/mo | None |
| ChatGPT Pro | from $100/mo (tiers at $100 / $200 / $500) | First Dot included |
| Business Premium | $125/user/mo ($100 billed annually) | First Dot included |
Notes that matter for budgeting:
- Your first Dot is bundled with Pro and Business Premium — there's no standalone Dot SKU yet
- OpenAI says you'll later be able to buy additional Dots and pay to increase a Dot's speed or monthly workload, but hasn't published those prices
- The Pro rollout currently excludes the EEA, Switzerland, and the UK; Business Premium is listed as available across all supported regions
So the realistic entry price for a sales team is $100–$125 per user per month — roughly what a seat of a mid-tier sales engagement platform costs.
Usage limits: the fine print
Dots are not unlimited workers. Current published limits:
- Up to 15 active tasks per Pro user
- Aggregate event limits of 30 runs per hour and 720 per day
- An allowance for "deeper work," with extended limits during the first month
For an individual AE managing their own follow-ups, that's plenty. For an SDR team hoping one Dot will run 500 personalized outreach tasks a day — it won't. That's an important constraint before anyone pitches Dots as an AI SDR replacement. (For tools purpose-built for that job, see our best AI BDR tools comparison.)
Everything else at DevDay 2026 that matters for GTM
GPT-6.1 Sol — a new model at $2 input / $10 output per million tokens, roughly one-fifth of Astra's price at near-Astra quality, with better coding and computer use. If you've built internal sales tooling on the API, this is your new default. OpenAI also claims up to 8x faster token generation in Codex and 6x in the API.
Agents API with computer use — the Agents API (public beta since September 10) got hosted browser access at DevDay. Developers get the Codex harness as a managed service: OpenAI handles sessions, orchestration, and context recovery; your agents can execute code, connect to MCP servers, and delegate to other agents. This is the build-your-own-Dot path for RevOps teams.
ChatGPT Space — a shared environment where teammates and Dots collaborate on files, presentations, spreadsheets, and project context. A new document type called Pages, built for human–agent collaboration, lands in Q4 2026. This is OpenAI's answer to the collaborative-workspace push we covered in Claude Docs and Slides for sales teams.
OpenAI Marketplace (beta) — eligible enterprise customers can apply part of their OpenAI spend commitment toward qualifying partner software. Worth a conversation with your OpenAI rep if you're already on an enterprise agreement.
5 Dot workflows for sales teams to test first
Keep the limits in mind (15 active tasks, 30 runs/hour) and start with high-leverage, low-frequency work:
1. Morning pipeline brief
Give a Dot read access to your CRM connector and Slack. Recurring task: every weekday at 7am, summarize deals that moved stages, meetings booked overnight, and contacts who went dark for 7+ days. Deliver to Slack. This is the same pattern we use in our daily SDR workflow with Claude — Dots just make the scheduling native.
2. Pre-meeting research packets
Dot watches your calendar. Two hours before any external meeting, it browses the prospect's site, recent funding news, and LinkedIn activity, then drops a one-page brief in your inbox. Its own browser means it can do this without your machine on — a real edge over local-only setups like the ones in our Codex prompts for SDR productivity guide.
3. Follow-up enforcement
Hand the Dot your sent folder. Task: if a prospect hasn't replied in 4 business days and there's no scheduled next step, draft a follow-up in your voice and queue it for approval. Keep a human on send — Dots are too new to give them your sender reputation.
4. Conference and event list building
Point a Dot at an event's public exhibitor directory and ask for a structured list of companies, booth numbers, and likely ICP fits. Its hosted browser handles the clicking. (If you need this at scale, a dedicated tool is still faster — our free conference scraper does it in minutes.)
5. Competitive watch
Weekly recurring task: check competitors' pricing pages, changelogs, and release notes; report only what changed. We've caught multiple silent price increases this year doing exactly this — vendors change pricing far more often than they announce it.
The catch: generalist agents still don't know your buyers
Here's the pattern we keep seeing — first with Claude Opus 5.5, then Salesforce inside Claude, now Dots: general-purpose agents are getting genuinely good at executing tasks you define. What they can't do is tell you which accounts are in-market right now and what to do about them.
A Dot will happily research any account you name. It has no idea that a VP of Sales from your ICP visited your pricing page twice yesterday, or that your champion at a target account just changed jobs. Signal detection — visitor identification, intent data, champion tracking — still has to come from somewhere, and that's the layer MarketBetter owns: it tells your team WHO is in-market and WHAT to do next, then a generalist agent (Dot, Claude, or your own OpenClaw agent) can execute the plays.
One more wrinkle: Dots browsing the web means even more of your buyers' research is now done by agents, not humans. Our own Search Console data shows AI agents are already searching Google with long persona-style prompts — Dots will accelerate that, and your content needs to be legible to them.
Should your sales team buy Pro for Dots?
Try it if: you're already paying for Pro or Business Premium, your reps lose hours to research and follow-up admin, and someone on the team will own prompt quality for the first month. The bundled first Dot makes the experiment effectively free.
Wait if: you're in the EEA/UK on Pro (not available yet), you need high-volume execution (the run limits will stop you), or you can't tolerate flaky behavior in revenue-critical paths — the demo stumble was a useful reminder that "always-on" is aspirational.
Either way: the direction is clear. Between Dots, the Agents API, and ChatGPT Space, OpenAI is turning ChatGPT into an operating system for delegated work — the same trajectory we're watching in the Claude vs ChatGPT for sales teams race.
Want the signal layer that tells your agents which accounts to work? MarketBetter identifies your anonymous website visitors, tracks champion job changes, and turns signals into specific next actions for your SDRs. Book a demo →