Two offers lower the entry threshold

OpenAI launched ChatGPT Team on 10 January as a self-service subscription for teams. It included a shared workspace, an admin console, higher usage limits and the ability to create and share customized GPTs. OpenAI said the organization controlled its data and that conversations were not used for model training. The announcement suggested uses such as project management, onboarding and data analysis, although these examples were vendor proposals and customer statements.

Microsoft announced on 15 January that Copilot for Microsoft 365 was available to more small and medium-sized businesses. It removed the 300-seat minimum, allowing Microsoft 365 Business Premium and Business Standard customers to buy between one and 299 seats. Copilot Pro also launched for individuals. Managers need to preserve the distinction: an organizational workspace, a licence attached to the company account and an individual subscription provide different administrative controls.

Project work needs shared rules

A lower purchasing threshold makes a pilot easier, but can distribute experiments without comparable measures. Choose two tasks in one project, such as drafting a status update and consolidating risks. Define approved source documents, the person who reviews the output and the location where the final result is stored.

Collaboration and productivity should be measured across the delivery: time to an approved result, number of corrections, reuse of shared instructions and quality of handovers. A fast first draft is not an improvement when the project manager must locate unsupported statements or reconcile several parallel versions.

Administration should follow the use case

Assign an owner to each shared AI configuration, record its instructions and knowledge sources and set a review date. Test access with ordinary user roles. Track where employees use personal subscriptions for company work because January's wider availability made that boundary more relevant.

The releases mark a shift from individual experiments toward workspaces that a team can administer. This creates better conditions for shared learning but does not by itself demonstrate higher quality. Each organization must show that AI support improves a bounded process without weakening traceability, professional review or collaboration.

A four-week pilot can begin with one week without AI to establish a baseline, followed by two weeks using shared instructions and a final review. Ask participants to mark outputs they use, edit or reject. The project manager can then compare elapsed time with the number of material corrections. An administrator should also monitor membership, shared GPTs and departed users. Combining workflow measures with administrative control provides stronger evidence than message counts or a broad survey asking whether the tool feels useful.

Sources

OpenAI, “Introducing ChatGPT Team,” 10 January 2024

Microsoft 365 Blog, “Expanding Copilot for Microsoft 365 to businesses of all sizes,” 15 January 2024

Join the discussion

Which experience from your own organisation is most relevant to this story? Share what you learned.