The Race to Sell AI 'Agents' to Companies Heats Up

In brief: The biggest technology companies are competing hard to sell AI 'agents' — software that does not just answer questions but carries out multi-step tasks — to large organisations. Reports point to Google positioning its enterprise agent platform around governance and control; OpenAI moving its workplace agents to credit-based billing (charging for the work agents actually perform); and Anthropic extending its Claude 'Cowork' product from desktop to web and mobile, with enterprise features aimed at compliance teams.

This article is general information about developments in legal technology.

What is an AI 'agent'?

In plain terms, an AI agent is an assistant given permission to use tools and take actions on your behalf — for example, gathering information across systems, preparing a document and filing it — with a human supervising rather than doing each step. For businesses, the appeal is doing more with the same staff; for providers, agents are a way to charge for outcomes, not just conversations.

Why is 'control' the big theme?

A recurring theme in these launches is control. Enterprise buyers, especially in regulated fields, do not want an assistant that quietly touches sensitive data. So vendors are competing on governance: permission settings, audit logs, and tools that let compliance teams monitor and restrict what an agent can see and do across chat, coding and office tasks.

Why it matters for law firms

For a law firm, this is directly relevant, because these are the very products a modern practice might use to draft, research and organise work. The opportunity is efficiency; the responsibility is oversight — keeping a human firmly in charge, confining agents to appropriate data, and confirming their output. What to watch is which platform wins over cautious enterprise and legal customers, and whether the governance controls prove strong enough to trust with confidential matters.

Frequently asked questions

What is an AI agent?

An AI assistant that is given permission to use tools and carry out multi-step tasks on your behalf, under human supervision — rather than simply answering questions.

Are AI agents safe for confidential legal work?

They can assist, but only with proper safeguards: strong governance controls, restricting the data an agent can access, keeping a human responsible for decisions, and confirming the agent's output. Confidentiality obligations still rest with the lawyer.

How should a firm start using them?

Begin with lower-risk tasks such as drafting and research, keep a qualified human in charge of every decision, protect client confidences, and verify results before relying on them.

Last updated: 6 October 2026.

This article is published by Khaw Ewe Seng & Co., Advocates & Solicitors (Penang) for general information about Malaysian law. It is not legal advice, and it does not create a solicitor–client relationship. For advice on a specific matter, please consult a qualified lawyer.