From Personal AI to Organisational AI
Most AI use in the workplace is still surprisingly individual. The work may be useful, but the intelligence behind it is largely private.
From individual tools to shared infrastructure
A lawyer opens ChatGPT or Claude, drafts a clause, researches a point, summarises a judgment or refines an email. The work may be useful, but the intelligence behind it is largely private.
The lawyer disappears into their preferred AI tool, does the work, then brings the output back to colleagues.
The Limit
That model is useful for one person. It is far less useful for the firm.
This is why Jack Dorsey’s Buzz is interesting.
A Shared Workspace
Buzz points towards a different interaction model, where humans and AI agents operate together inside the same shared workspace. Instead of each employee running their own isolated AI conversations, teams can see, use and improve the same agents together.
For a law firm, imagine a matter room containing:
The research agent finds relevant authorities. The case law agent challenges whether they are applicable. The drafting agent incorporates them into an opinion. The partner corrects the legal strategy. The QA agent checks the final draft against the evidence and instructions.
Crucially, the whole team can see what happened.
Firm Infrastructure
That changes AI from a personal productivity tool into a piece of shared firm infrastructure.
If one partner develops an excellent approach to analysing shareholder disputes, that capability no longer needs to live inside their private prompts. It can become part of a shared agent that associates across the firm can use.
Compounding
When someone improves the agent, everyone benefits.
Over time, the firm starts encoding its own ways of working, precedents, drafting standards, legal reasoning processes and quality controls into a collection of reusable agents.
The shift looks something like this:
Today
50 lawyers × hundreds of isolated AI conversations.
Tomorrow
50 lawyers + a shared team of specialised agents + continuously improving firm knowledge.
That is the important stepchange.
Slack made workplace communication communal. GitHub made software development communal. Platforms like Buzz suggest that AI itself may now become communal.
For professional services firms, that may prove far more important than simply giving every employee access to ChatGPT.
The competitive advantage will not necessarily be who has the smartest individual AI users.
It may be who builds the smartest human-agent organisation.