OpenAI Launches ChatGPT for Sales With Salesforce and HubSpot
Key Takeaways
- OpenAI introduced ChatGPT for Sales on 17 November 2026, pairing a new OpenAI-built Sales plugin with more than 30 new and enhanced partner plugins, including Salesforce, HubSpot, Close CRM, and monday CRM.
- The product brings customer records, call notes, and sales materials into ChatGPT, so sellers can do account research, meeting preparation, follow-up drafting, and pipeline reviews without leaving the chat.
- Partner plugins also cover prospecting (Apollo, Clay, Common Room), call notes (Gong, Otter.ai, Granola), and quotes and agreements (Docusign, Ironclad, PandaDoc).
- Work can be handed to a dot, OpenAI's always-on agent in ChatGPT, which can use the same plugins to prepare meeting briefs, track deal updates, and propose next steps for review.
- ChatGPT Business and Enterprise administrators enable the Sales and partner plugins through workspace Plugins controls; availability varies by plan, region, and workspace settings, and some connections require administrator setup or approval.
- OpenAI has not announced a separate price for ChatGPT for Sales itself; the underlying ChatGPT Business plan costs $25 per user per month, or $20 per user per month on annual billing, with a two-seat minimum.
What OpenAI Announced
On 17 November 2026, OpenAI launched ChatGPT for Sales, a package aimed squarely at the daily work of sales teams. The centrepiece is a new Sales plugin built by OpenAI itself, combined with ready-made seller workflows and more than 30 new and enhanced partner plugins. The idea is simple to state and ambitious in practice: instead of ChatGPT being a separate tool that a salesperson opens to draft an email, it becomes the place where the selling itself happens, with the customer data already inside it.
The announcement was reported on 17 November, the same day OpenAI made its cybersecurity models available to NATO, and it fits a clear pattern in the company's recent moves. OpenAI is no longer selling ChatGPT mainly as a chatbot for individuals. It is building it into a work platform for businesses, one department at a time. Sales is the latest department to get its own version, and it is one of the largest.
The product page, as reported, says teams can capture customer insights and proven sales plays in shared skills and ChatGPT Sites, so that what one good seller learns can be reused across a revenue organisation. That is a notable detail, because it shows OpenAI thinking about sales knowledge as a shared asset rather than a set of individual tricks. A playbook that lives inside ChatGPT, and that every rep can draw on, is a different proposition from a document sitting in a shared drive.
What the Sales Plugin Does
The Sales plugin covers the everyday work that fills a seller's day. The workflows OpenAI described include deciding which accounts to pursue, preparing for a call, and drafting the follow-up afterwards. Account research, meeting preparation, follow-up, and pipeline reviews are the four jobs named most often in reporting of the launch.
Each of those jobs is worth unpacking, because the value of the product depends on how well it does them. Account research means pulling together what is known about a customer, from the CRM record to recent call notes, and turning it into a brief a rep can read in minutes. Meeting preparation means going a step further: not just summarising the account, but suggesting what to ask, what to raise, and what to avoid. Follow-up means drafting the email or message after the meeting, grounded in what was actually discussed rather than in a generic template. Pipeline review means looking across open deals, spotting the ones that have stalled, and flagging the risks.
None of these jobs is new. Salespeople have been doing them for decades, and there are already many tools that help with each one. What is new is the claim that they can all be done in one place, in plain language, with the data pulled in automatically. Whether that claim holds up in practice will depend on the quality of the CRM connections, which is where the partner plugins come in.
Salesforce, HubSpot, and the Partner Ecosystem
The partner ecosystem is what makes ChatGPT for Sales more than a drafting tool. Salesforce, HubSpot, Close CRM, and monday CRM can contribute account and deal information, while Apollo, Clay, and Common Room add prospecting signals. For the later stages of a deal, Gong, Otter.ai, and Granola bring in call notes, and Docusign, Ironclad, and PandaDoc cover quotes and agreements.
Salesforce and HubSpot deserve particular attention, because they are the two CRM systems most sales teams actually use. The Salesforce connection runs through OpenAI's Salesforce plugin, which includes Salesforce's Agentforce Sales app and can read and write records, including reviewing records, summarising pipeline risk, and preparing follow-ups and CRM updates. HubSpot's connector works on all HubSpot products and plans: it reads contacts, companies, deals, tickets, and activities, and it can create and update records, up to 10 at a time, with no deletes. HubSpot recommends setting write tools to Needs Approval, which is a sensible safeguard when an AI system can change your customer database.
That last point matters. A plugin that can write to a CRM is powerful, but it is also a risk. A wrong update, a duplicated record, or a note written to the wrong account can cause real confusion. The fact that HubSpot's connector has a no-deletes rule, and that write actions can be set to require approval, shows that the vendors have thought about this. Administrators would be wise to keep those safeguards on, at least at first, and to treat the AI's CRM writes as drafts until the team trusts them.
The breadth of the ecosystem is also a signal about OpenAI's strategy. Rather than building every integration itself, OpenAI is letting the specialist vendors build theirs, and providing the platform they plug into. That is how ChatGPT becomes a common interface for work spread across several tools, as the launch coverage described it, rather than just another tab in the browser.
Dots: The Always-On Sales Assistant
The other half of the announcement concerns dots, OpenAI's always-on agents in ChatGPT, which launched at DevDay on 29 September 2026. A dot runs on its own cloud computer and keeps working between conversations, which makes it a natural fit for sales work that happens on a schedule rather than on demand.
Users can hand work to a dot, which can use the same plugins to prepare meeting briefs, track deal updates, and propose next steps for review. The key phrase there is "for review". The dot proposes; the human decides. That is the right model for sales, where a wrong move can cost a deal, but it also raises a practical question: how much time does a rep save if they still have to check everything the dot does? The honest answer is that it depends on how often the dot gets it right, and that will only become clear once teams have used it for a few months.
Dots are not cheap. A dot costs at least $100 a month per rep on the Pro plan, or a Business Premium seat at $125 a month, $100 billed annually, with a two-seat minimum per workspace. That pricing, reported in October 2026, means dots are a serious commitment for a sales team, not a casual add-on. OpenAI said it is giving dots extended limits for the first month and will publish usage terms for each plan after that, so the economics may change. But for now, a sales leader considering dots needs to weigh a real per-rep cost against a still-unproven time saving.
Who Gets It, and How It Is Switched On
ChatGPT for Sales is a business product, not a consumer one. ChatGPT Business and Enterprise administrators can enable the Sales plugin and chosen partner plugins from the admin console, through workspace Plugins controls. Availability of partner plugins and features varies by plan, region, and workspace settings, and some connections require administrator setup or approval.
That administrative gate is important, because it means ChatGPT for Sales will not simply appear for every ChatGPT user. It has to be switched on, deliberately, by someone responsible for the workspace, and the plugins have to be chosen and configured. That is the right way to roll out a tool that can read and write customer data, but it also means adoption will be gradual. The companies that benefit first will be the ones whose administrators move quickly and whose sales teams are willing to change how they work.
It is also worth noting what OpenAI has not said. The company has not announced a separate price for ChatGPT for Sales itself, and reporting of the launch does not include a standalone price. The underlying ChatGPT Business plan costs $25 per user per month, or $20 per user per month on annual billing, with a two-seat minimum, according to OpenAI's pricing as checked in October 2026. Enterprise pricing is quoted by sales. Whether the Sales plugin and its partner integrations carry an additional charge, and whether some partner plugins require their own subscriptions, has not been disclosed, and buyers should confirm the full cost before committing.
The Evidence So Far: One Early Number
The launch coverage includes one early user estimate that is worth reporting with its limits stated plainly. NeoBid co-founder Gunnar Gombert estimated that using HubSpot with ChatGPT for Sales reduced manual effort by around 65% across prospect research, CRM enrichment, opportunity prioritisation, and personalised outreach. The estimate covers those activities rather than the entire sales cycle, and it comes from a single early user, not from a controlled study.
That is the kind of figure that should be treated as a promising anecdote, not as a result. A 65% reduction in manual effort on four specific tasks is plausible for a tool that automates research and drafting, but it says nothing about whether deals close faster, whether revenue rises, or whether the time saved is spent on selling rather than on checking the AI's work. OpenAI has not published its own productivity figures for the product, and no independent evaluation exists yet. Sales leaders should ask for a trial, measure the time saved on their own team's real work, and decide on that evidence rather than on a launch-day estimate.
How This Fits OpenAI's Business Push
ChatGPT for Sales does not arrive in isolation. It is the latest step in a deliberate push by OpenAI into business software, and the pattern of the last few weeks makes that clear. On 12 November 2026, OpenAI's ChatGPT Business Hub was in the news, and we covered what the ChatGPT Business Hub is and how it works as the company began organising its business offering more clearly. The OpenAI and ServiceNow partnership around ChatGPT and Codex showed the same direction: ChatGPT moving inside the enterprise software stack, not sitting beside it. And the launch of dots in September gave OpenAI the always-on agent that a sales product needs.
The logic is straightforward. OpenAI's consumer business is large, but its future growth depends on businesses paying per seat, per month, for tools their employees use every day. Sales is one of the largest such markets, and it is one where the work, research, writing, preparation, follow-up, is exactly the kind of work language models do well. If ChatGPT can capture a share of the sales workflow, it becomes much harder for a company to cancel, and much easier for OpenAI to charge for.
There is also a competitive logic. Salesforce has its own AI, Agentforce, and HubSpot has built AI throughout its platform. By making Salesforce and HubSpot plugins for ChatGPT, OpenAI is positioning itself as the layer above the CRM, the place where the seller works, rather than the system of record underneath. That is a comfortable position if it works, because it means OpenAI does not have to replace the CRM. But it also means OpenAI depends on those vendors continuing to support the integration, and on sellers preferring to work in ChatGPT rather than inside the CRM itself.
What Sales Teams Should Watch
For a sales leader considering ChatGPT for Sales, the launch raises four practical questions.
First, data quality. The product is only as good as the CRM data it reads. A team with clean, complete, up-to-date records will get useful briefs and accurate pipeline reviews. A team with patchy records will get confident-sounding summaries built on gaps, and the AI will not always flag what is missing. Before switching the product on, it is worth auditing the CRM.
Second, write permissions. The ability to update records is powerful, but it should be rolled out carefully. Start with read-only access, let the team see what the AI suggests, and only enable writes, with approval required, once the suggestions are consistently reliable. HubSpot's recommendation to set write tools to Needs Approval is a good default.
Third, measurement. The 65% figure from one early user is not a benchmark. A team should pick a small set of measures, such as time spent on meeting preparation, or the share of follow-ups sent within a day, and track them before and after adoption. If the numbers do not move, the product is not earning its cost, however impressive it feels.
Fourth, cost. With no separate price announced for the Sales product, and dots priced from $100 a month per rep, the total cost of a full rollout could be substantial. Sales leaders should get a written quote that includes the Business or Enterprise seats, any Sales-specific charges, and the cost of the partner plugins their team needs, before committing.
For a broader view of how AI is changing the way teams sell, our guide to AI tools for sales and how modern teams sell smarter covers the wider market that ChatGPT for Sales is entering, including the tools it will compete with and complement.
What We Could Not Verify
- A separate price for ChatGPT for Sales itself. OpenAI has not announced one in the reporting we reviewed, and the Business plan price quoted is for the underlying ChatGPT Business plan.
- Whether every partner plugin is available in every region and on every plan. OpenAI says availability varies by plan, region, and workspace settings, and no full availability list has been published.
- The 65% manual-effort reduction estimated by one early user. It is a single-user estimate covering four specific activities, not an independent or controlled measurement.
- The full list of the more than 30 partner plugins. Reporting names the main categories and vendors, but a complete list has not been published in the sources we reviewed.
- The contents of the Reuters report on the launch. We were unable to retrieve the Reuters article directly; the launch details in this article are drawn from OpenAI's announcement as reported by Unite.AI and AI Brief on 17 November 2026.