ChatGPT is moving closer to the tools your team already uses

A conversation closer to the work

Finding an answer often means opening several tools, checking a record, and asking someone what happens next. This week's OpenAI announcements with Oracle and ServiceNow point to more of that work becoming possible through ChatGPT.

The October 27 Oracle announcement describes plans to bring business applications and information together in ChatGPT through Oracle AI Workspace. On October 28, OpenAI and ServiceNow described expanded connections for investigating requests and taking approved actions. These are partnerships with different products and timelines, so it helps to separate what is available from what is planned.

Read the announcement: OpenAI expands partnership with Oracle to bring its apps to ChatGPT.

Check the part you can use today

ServiceNow's existing ChatGPT plugin is available now. The announcement expects enhanced capabilities by the end of 2026, while the planned Codex connection for building ServiceNow apps is expected in 2027. The proposed experience includes finding related incidents and updating a record while keeping the company's rules and permissions in place.

OpenAI describes the broader Oracle AI Workspace experience as being developed. Its example brings together information about costs, purchasing, and unfinished month-end work. Separately, Oracle customers in the US can already purchase OpenAI tools through Oracle Marketplace. Check the particular application, access requirements, and release stage before making plans for your team.

Follow one request from beginning to end

Try this with a colleague who regularly handles an internal request. Ask them to walk you through one completed example. Where did the request arrive? Which record did they check? Who needed to approve the next step?

Notice where the person simply needed help finding information, and where they had to make a decision. A useful AI connection should respect that difference. Finding a related record and changing it are different responsibilities.

This gives you a practical way to explore the announcements without connecting every system at once. You can choose one helpful action and describe exactly what a person should see before approving it.

1. Choose a completed request you have permission to discuss.
2. List the tools and records the person used to finish it.
3. Mark each step as finding information, drafting, deciding, or changing something.
4. Pick one step to explore and agree who checks the result.

Make the connection useful to your team

Our view is that these partnerships are worth watching because they bring AI closer to information people already depend on. The value will come from making a particular job easier while keeping the right person in control.

A guided walkthrough can help you learn how to ask for that help and check the answer. For a custom tool, the same completed request gives us something concrete to explore together. We can start with how the work happens today and what you would like to become easier.