Recently, many legal departments have begun using artificial intelligence to summarize contracts, conduct research, prepare drafts, and answer questions. The next step is more significant: AI is moving beyond simply providing answers and is now beginning to execute processes.

That, in simple terms, is what Agent AI proposes: systems capable of receiving a goal, dividing it into tasks, using available information, interacting with other tools, and advancing in a workflow with much less human intervention.

For an in-house legal department, the possibilities are concrete. One agent could receive a contract, identify clauses outside the playbook, compare them to previously approved positions, prepare draft wording, identify who needs to approve an exception, and follow up until closing. Another could monitor regulatory changes and alert only to those that potentially affect the company.

The difference is important: it’s not just about doing legal work faster, but about reviewing how legal services are organized and provided within the company.

The first step is not to buy AI

Before implementing an agent, the department should ask itself where it is actually wasting time.

Mapping the main workflows often reveals repetitive activities:intake of requests, initial review of contracts, NDAs, tracking of contractual obligations, compliance questionnaires, reports, document management and regulatory monitoring.

A good rule of thumb is to separate three categories: what can be automated, what requires legal judgment, and what requires a business decision.

AI is an extraordinary tool and can radically increase the efficiency of a legal department. But just because it can produce a response or perform a task doesn’t mean we should stop reviewing it.

If the process is poorly designed, AI will simply automate the chaos. Before automating, processes must be standardized, rules defined, reliable sources established, and it determined when an issue should be escalated to a lawyer.

“Hallucinations” remain a problem

Reliability continues to be a major concern.

Models can produce a legally convincing but incorrect response: a nonexistent judgment, an interpretation that does not arise from the law, or a quotation that appears authentic but is not. Academic research on AI-based legal tools has shown that this risk persists even in specialized systems.

Therefore, a rule should be part of any internal AI policy: the AI outcome is a starting point, not the legal authority.

In practice, this means verifying the original source, confirming jurisdiction and validity, and checking if the conclusion actually corresponds to the text.

And the greater the risk, the greater the level of review should be.

This is especially important with Agentic AI. An incorrect response from a chatbot can lead to a poor recommendation. An agent connected to the contracts, email, or compliance system could also perform an incorrect action.

Therefore, greater autonomy should not mean less control, but rather better-designed control.

The lawyer who knows the business

One of the most interesting changes will be in the role of the In-House lawyer.

A significant part of the traditional work of junior lawyers consists of activities that AI can perform increasingly better: initial document review, legal research, contract comparison, due diligence, and first drafts.

This doesn’t necessarily mean that junior lawyers will disappear. It does mean that the traditional training model will have to change. If AI performs all the basic work, a young lawyer may lose some of the opportunities that historically allowed them to develop legal judgment.

But there is another possibility. If technology allows for a significant reduction in the time spent on repetitive tasks, the lawyer can dedicate more time to understanding the company.

In my view, in-house counsel must be increasingly involved with sales, product, operations, finance, technology, marketing, procurement, and strategy. This is not only to understand their legal needs, but also to grasp how the business truly operates, where value is generated, and where revenue is generated.

A lawyer who understands the contract can provide a sound legal response. One who also understands the client, the product, the business process, the risks, and the company’s objectives can contribute to a much more comprehensive business decision.

AI shouldn’t take lawyers away from the business. It should give them more time to be in the business.

El nuevo rol del In-House Counsel

For legal leaders, the question should no longer be simply: “Can we use AI?”

The important questions are different: What processes do we want to automate? What decisions should never be made autonomously? What sources can the system use? Who verifies the results? When should a lawyer be involved? What information can be used?

A mature department should establish clear rules regarding confidentiality, authorized tools, sources of information, approval levels, documentation, and escalation.

It should also measure the outcome: response time, volume of issues handled, reduction of repetitive tasks, compliance with SLAs and, especially, impact on the business.

The goal is not to eliminate the lawyer from the process. It’s to place the lawyer where their judgment adds the most value.

AI can research, compare, summarize, propose, classify, and increasingly, execute. The lawyer must continue to determine when the result is sufficient, when it needs to be verified, and when a decision requires legal and business judgment.

The future of in-house legal will likely involve less time spent searching for information, copying data, and reviewing repetitive documents, and more time spent on negotiation, strategy, risk management, and advising the business.

The real transformation won’t be AI replacing the legal department. It will be that departments that know how to integrate people, processes, data, and AI will be able to do much more with the same resources.

True success will not be having the most automated Legal department.

It will mean having a Legal department that, thanks to technology, has a greater capacity to understand the business, participate in its decisions, and directly help generate value for the company.

That, ultimately, is the true potential of an AI-enabled Legal Function: using technology to be more efficient, be closer to the business and, above all, provide more value where it is actually generated.