AI in salesThomas Honemeyer, Founder and Managing Director
ChatGPT for sales: what the new AI assistants from OpenAI and Microsoft mean for retailers
Microsoft and Salesforce bring customer data into Copilot, and OpenAI is aiming ChatGPT specifically at sales teams. What the artificial intelligence (AI) assistants can do, who they are for and which questions businesses with regular customers should draw from them.
Key points at a glance
- On 17 November 2026, Microsoft and Salesforce presented the Copilotforce partnership, which brings customer data from Salesforce into Microsoft Copilot and Teams.
- With ChatGPT for Sales, OpenAI offers its own product for sales teams that connects customer data from systems such as Salesforce, HubSpot and Gmail.
- Both assistants prioritise customers, prepare meetings and draft next steps, with a person confirming changes.
- The offers are aimed mainly at business to business sales with a well maintained customer relationship management system, not at caring for regular customers in a shop.
- For businesses with regular customers, other data, other occasions and a personal channel such as WhatsApp count.
What did Microsoft and Salesforce announce on 17 November 2026?
Anyone searching for ChatGPT for sales has so far mostly found collections of prompts. Two announcements are now changing the picture: on 17 November 2026, at the Microsoft Ignite conference, Microsoft and Salesforce presented the Copilotforce partnership. OpenAI, too, is addressing sales teams with its own offer, ChatGPT for Sales. Both approaches bring customer data from customer relationship management (CRM) directly into the assistant employees already work with.
According to Microsoft, Copilotforce includes:
- Salesforce in Copilot: an extension with more than 38 prepared sales skills, for example for reviewing open opportunities, deal planning, account health, upcoming renewals and follow up after meetings. It has been available as a public preview since 17 November 2026.
- Context from everyday work: the assistant includes Teams meetings, Outlook emails and documents, so that customer data and conversation history are visible together.
- Sales Cloud in Teams: opportunities, target achievement and deal insights appear directly in the Teams chat. This feature is currently a pilot with large customers.
- Confirmation by a person: according to Microsoft, the assistant only writes changes to the CRM after explicit confirmation by the user.
Microsoft does not mention prices in the announcement. According to Microsoft, existing Salesforce permissions remain in place: an employee only sees in the assistant what he or she may also see in the CRM.
What does OpenAI offer with ChatGPT for Sales?
According to OpenAI, ChatGPT connects customer data from systems such as Salesforce, HubSpot and Gmail through extensions. A dedicated sales module provides prepared workflows for typical tasks: prioritising customers, preparing meetings and spotting risks earlier. According to OpenAI, AI agents handle research, follow up and CRM updates, while the team stays in control. OpenAI does not state prices or regional availability on its solution page.
What do both announcements have in common?
- The customer data comes to the assistant instead of employees switching between programmes.
- The assistant prioritises, prepares and drafts. A person decides.
- Both offers are designed for sales teams with opportunities, quotes and deals, so primarily for business to business sales with a well maintained CRM.
The question in sales is therefore shifting: it is less about whether a team uses AI and more about whether the assistant knows the right data and derives a concrete next step from it.
Does such an assistant suit a business with regular customers?
A jeweller, a car dealership, a fashion house or a hotel works differently from a business to business sales team. There is rarely a list of open deals with dates. Revenue comes when a regular customer returns. The data for this is usually in the till system, inventory management, the dealer management system (DMS) or the hotel software, not necessarily in Salesforce or HubSpot. And contact is personal: in the shop, by phone or via WhatsApp.
Three questions help to assess an AI assistant for sales:
- Where is your customer data? Can the assistant read it without your team maintaining data twice?
- Does the assistant recognise the occasions that matter for your customers? For example a service that is due, an anniversary or new arrivals that match an earlier purchase. How this works is shown on the page on customer data analysis with AI.
- Who approves a message? A personal tone and approval by a person protect the trust of your regular customers. Why a single message with a genuine occasion achieves more than a broadcast to everyone is explained under personal WhatsApp marketing.
What should decision makers review now?
- Data basis: which information about your customers is available today, such as purchases, appointments, preferences and consents, and how up to date is it?
- Everyday work: how much time does your team have each day for contacting customers? An assistant helps if it directs this time to the customers for whom contact makes sense today.
- Channel: how do your employees actually reach customers? An assistant that only drafts emails does not suit a business that writes to regular customers via WhatsApp.
- Data protection: clarify with your data protection officer where customer data is processed and on what basis you contact customers. This article does not replace legal advice.
Frequently asked questions
What is Copilotforce?
Copilotforce is a partnership between Microsoft and Salesforce announced on 17 November 2026. It brings customer data and sales workflows from Salesforce into Microsoft Copilot and Teams. According to Microsoft, the Salesforce in Copilot extension has been available as a public preview since 17 November 2026.
What can ChatGPT do in sales according to OpenAI?
According to OpenAI, ChatGPT connects customer data from systems such as Salesforce, HubSpot and Gmail through extensions. It helps prioritise customers, prepare meetings, spot risks and follow up, while the team stays in control.
Are these assistants suitable for retail with regular customers?
Both offers are designed mainly for business to business sales with a well maintained CRM. Businesses with regular customers should check whether the assistant can read their data from till, inventory or dealer management system, whether it recognises suitable occasions and who approves messages to customers.