A new era of artificial intelligence with agentic AI

Unlike generative AI models like ChatGPT or Gemini, which need input, agentic AI actively seeks out objectives. It doesn’t merely assist; it initiates and executes.

Agentic AI may be the missing link between a good chatbot and a real life colleague: a solution that finally acts as a “helping hand” – something all law firms are keen to obtain.

Few lawyers have taken the time to understand enough about how generative AI models function and the appropriate use of these models, including steps that would mitigate hallucinations – which is when AI gives incorrect or incomplete results – or other risks and issues that arise.

How agentic AI works

Agentic AI refers to systems that can interpret goals, plan strategies, carry out tasks and revise actions with little or no human input.

Unlike traditional AI, which responds to prompts, agentic AI works toward long-term objectives and adapts as it goes.

To understand the difference, consider generative AI as the intellect and agentic AI as both the mind and hands. Where a model like ChatGPT might help draft a report when asked, an agentic system can plan, write, revise and deliver it without further instruction.

Real world examples

Consider the example of acting for a buyer in a private equity transaction.

The job typically begins with drafting a Non-Disclosure Agreement (NDA), followed by reviewing documents during due diligence.

A law firm’s role is then to assist our client in negotiating the sale and purchase agreement, with each step involving multiple rounds.

Properly configured agentic AI could be assigned to: “Support the legal team through all stages of the transaction to completion.”

The system would immediately break down this objective into individual tasks, such as identifying transaction milestones and drafting the NDA using templates aligned with the firm’s preferred clauses.

To highlight another example, if the seller uploads a vendor due diligence report containing ambiguous language, generative AI might highlight the section upon request.

In comparison, agentic AI autonomously identifies the inconsistency, cross-references it against previously flagged concerns in the same sector, consults internal deal databases for how similar issues were handled and proposes three clause amendments.

It also generates a short internal memo summarising the issue, complete with a risk assessment and recommended approach.

Agentic AI, data and agents

In this scenario, the AI doesn’t just answer questions – it acts like an assistant, handling background tasks while lawyers focus on strategy and client relations.

Agentic AI data comes from various sources. These include publicly available datasets, proprietary databases, user interactions and real-time inputs.

This data is the foundation agentic AI uses to make decisions, plan next steps and adapt strategies over time without human intervention, although at the moment a human element is required.

The distinction between agentic AI and AI agents can also be confusing. Agentic AI refers to systems that demonstrate autonomy. AI agents are the parts of these systems that carry out actions.

Many AI agents aren’t truly agentic and operate within strict boundaries, like Siri or Alexa.

As integration and reasoning capacity improve, more agents are becoming able to operate across tasks and domains. This raises questions about superintelligence – AI systems that might eventually exceed human knowledge.

While we are not there yet, agentic AI represents a substantial step in that direction.

What agentic AI could mean for law firms

The implications of agentic AI in the workplace could be far-reaching, meaning efficiency gains, lower costs and increased productivity.

Junior lawyers, analysts and administrators may see parts of their jobs handed over to intelligent systems. This will require a shift in education.

Professionals must understand how these systems work, and a good grasp of data, logic and AI tools are likely to become essential skills.

Risks and rewards

However, risks accompany the rewards.

When machines act independently, questions of accountability and ethical boundaries arise, such as who is responsible when an autonomous system makes a flawed decision.

These questions are central to the responsible use of AI. It’s vital for firms and users to have appropriate training in place, along with policy frameworks and an understanding of where the systems can be best used.

With many innovations, significant gaps emerge between those organisations that are early adopters and those who are fearful of embracing new systems and fail to understand them properly, falling behind as they do so.

Next steps to agentic AI

Law firms that want to benefit from agentic AI should ensure a good understanding of AI across all levels of the organisation, from senior partners to trainees and support staff.

Understanding how agentic AI systems function, where they excel and where their limitations lie is vital.

Becoming AI-ready isn’t simply about acquiring the latest tools, it’s about reshaping workflows, governance structures and staff mindsets to accommodate a new kind of digital colleague.

The future of law will not just reward firms that use agentic AI, but those that understand it, manage it responsibly and integrate it intelligently into the practice of law.