From GPT to Agents

Since ChatGPT first captured our imagination, AI has amazed us with its ability to generate human-like text, unravel complex queries, and be a copilot for our daily tasks. But that was just the beginning. 

Now, we’re entering a new era – the era of AI agents. 

Agents have begun to pop up everywhere, from customer support to retail, and now law. The global market for AI agents is booming: it is expected to balloon to a staggering $52 billion of revenue by 2030,¹ nearly 10x the $5.7 billion in 2024. Adoption of AI by the legal and professional services industries has also skyrocketed – with an estimated 72% of companies already actively deploying AI in their daily business functions.²

In 2025, the pace of AI acceleration is only intensifying, with over two thirds of executives actively mapping out agentic workflows.³  

But what exactly are AI agents, and why are they suddenly at the heart of innovation in law?

Defining AI Agents 

An AI agent⁴ is a system that can forwardly plan, collaborate, and take actions to accomplish specific goals. At their core, agents provide a more dynamic and capable framework for accomplishing complex tasks than traditional model-based systems.

Unlike basic automation tools that follow rigid rules or traditional chatbots, AI agents can:

• Understand context and nuance in complex information;

• Adapt to new situations as the environment changes;

• Handle multi-step processes autonomously; and

• Learn from feedback.

Think of an AI agent less like a conversational chatbot that passively responds to commands, and more like a digital assistant that proactively anticipates needs and dynamically takes initiative on workflows.

Why now? Recent Technological Breakthroughs 

Recent leaps⁵ in the capabilities of foundation models have made AI agents that are truly useful possible. 

Chain-of-Thought Reasoning 

The latest large language models (LLMs) from AI labs like OpenAI and Anthropic can now follow complex chains of reasoning. This allows them to break down problems into logical steps and arrive at sound conclusions - something previous AI models could not do reliably.

For example, when analyzing whether an invention is patentable, these reasoning models can consider multiple documents and formulate a strategy before responding. This allows AI agents to break down the task into discrete, sequential steps:

• Identify the key inventive features;

• Compare them systematically to prior art;

• Determine whether differences would be obvious to a skilled practitioner; and

• Formulate legal arguments based on recent case law.

Multi-modal Understanding

These cutting edge AI systems can now process text, images, diagrams, and code all together. This is crucial for patent work, where understanding technical drawings alongside written descriptions is essential.

Dynamic Tool-Use

Modern AI agents can call upon external tools when needed - searching databases, analyzing documents, or generating visualizations. This allows them to handle complex patent workflows that require multiple tooling capabilities.

Custom Agents for Law

Patent law presents unique challenges that make it particularly suitable for AI agents, such as information density (there are millions of existing patents, technical literature, highly technical legal precedents) and process complexity (each workflow involves dozens of interconnected steps). 

Static Tools to Dynamic Workflows

Here’s how we see AI agents enhancing the workflows of patent lawyers: 

• Searching: Traditional approaches miss approximately 30% of relevant prior art. AI agents have the ability to autonomously search the entire internet for patents, NPL, and products, while deeply understanding subtle technical connections and goals.

• Reviewing: With advances in the ability of models to reason about work, AI agents are highly skilled at and better suited to reviewing large amounts of complex, technical drafts for errors, consistency, and more.

• Editing: Generative AI approaches are strong at making first drafts of content. Agents extend this to excellence in editing – they are highly proficient in understanding human feedback in order to make multi-file systematic changes to content, ideas, and formatting. 

What’s next?

A Collaborative Future: Attorneys & AI Agents

What does the future of patent law look like? Crucially, AI agents will not replace humans, but augment them. AI agents work best when paired with attorney oversight — agents handle tedious, manual parts of the workflow, freeing lawyers to focus on applying their nuance, expertise, judgment from years of deep experience to solving complex legal problems in the patent space. 

Agents as a Hidden Superpower 

A Forbes study⁶ found that lawyers who effectively utilize AI completed tasks 50% faster and with 28% higher quality than their peers who did not. Agentic tools will widen this gap still further. As AI legal tools become the industry standard, forward-looking firms who embrace the advanced capabilities of agents will enjoy an early advantage in delivering deeper insights and better client outcomes.  

For example, &AI’s industry-leading AI agents are able to complete prior art searches and draft claim charts in a fraction of the time it used to take. In addition, they are able to complete multiple rounds of follow up searching, chart reviewing, and editing – amplifying time and cost savings even further. This allows firms to complete client work more effectively and efficiently, unlocking their ability to optimize fixed-fee workflows and allow teams to manage more projects at scale. 

Division of Labour in the Agentic Era

• Humans: Strategic thinking, creativity, ethics, client relationships.

• Agents: Data processing, searching, drafting, reviewing, editing.

The Road Ahead

At &AI, we believe that the future of patent law is hybrid. Lawyers will work hand-in-hand with AI agents to move faster, make better decisions, and deliver greater value to clients. 

This is why we’re developing novel professional-grade AI agents custom-built for patent practice. Our tools don’t just draft and search — they agentically reason, cite, and improve.

At &AI, our technological solutions are built with lawyers, for lawyers — designed from day zero to operate in close collaboration with legal professionals.

The GPT era gave us knowledge. In the agentic era, &AI will deliver outcomes.

We're just getting started.

¹ Reuters, AI Agents have Clear Mission, Feb 2025, citing a study by Boston Consulting Group (“BCG”).

² McKinsey, Why Agents are the Next Frontier of Generative AI, July 2024.³

⁴ BCG, BCG AI Radar, Jan 2025.

⁵ OpenAI, New Tools for Building Agents, Mar 2025; Anthropic, Building effective agents, Dec 2024.

⁶ Forbes, Lawyers Using AI Produce Better Work in Half the Time, Mar 2025.