The Next Frontier of AI: Moving Beyond Chatbots with Large Action Models (LAMs)

Generative AI is transforming industries at an unprecedented rate. Chatbots, virtual assistants, and co-pilots have become ubiquitous, enhancing everything from customer support to complex business workflows. But despite these advances, a question remains: Is this the full extent of AI’s potential, or are we merely scratching the surface?

According to a recent Goldman Sachs report, generative AI has yet to find its true “killer app”—the breakthrough innovation that drives mass adoption. Chatbots and co-pilots, while useful, are limited to one-way interactions and guided tasks. But what if AI could do more than just respond to commands? What if it could take action?

Enter Large Action Models (LAMs)—the future of AI technology that promises to transform the way we interact with machines, software, and even the world around us.

What Makes LAMs Different?

Large Action Models don’t just interpret language—they act on it. By utilizing existing programmatic pathways like APIs, or interacting directly with software interfaces, LAMs automate actions that would otherwise require human intervention.

For example, 3Pillar is working with clients to develop LAMs that automate repetitive business tasks, like data entry and customer outreach, allowing employees to focus on higher-level work. This is a step beyond traditional AI co-pilots, which still require human guidance. LAMs, in contrast, operate autonomously, learning from previous actions to improve their performance over time.

Even tech giants like Amazon and Salesforce are exploring LAMs as the next step in automation. For instance, Amazon’s semi-autonomous agents have already saved the company thousands of developer-years in maintaining Java code—an example of how AI agents can optimize business processes.

The Road Ahead: Challenges and Opportunities

Despite the excitement, the journey toward widespread adoption of LAMs isn’t without challenges. Unlike deterministic systems (traditional, rule-based programming), LLMs are probabilistic, meaning they can sometimes make unexpected or inaccurate decisions—known in AI circles as "hallucinations."

To mitigate this, developers are combining deterministic programming with LLMs, ensuring that the AI stays within predefined boundaries while still leveraging the power of language models. For example, 3Pillar is developing LAM applications that rely on knowledge graphs to keep responses focused, relevant, and legally sound. This ensures the AI doesn’t drift off-topic or suggest impractical solutions.

However, back-office applications, which deal with structured data and lower risk, provide an excellent testing ground for LAMs. As LAMs become more refined and reliable, we can expect them to take on more complex, high-stakes tasks in the coming years.

The Next 2-5 Years: AI Agents Will Transform Work as We Know It

What’s exciting is that this is just the beginning. In the next 2-5 years, we’ll likely see AI agents and LAMs becoming integral parts of our work and personal lives. According to a McKinsey report, AI agents are poised to power the next generation of automation—handling tasks like loan underwriting, code modernization, and marketing campaigns.

As companies invest more in developing LAM frameworks and integrating these systems into their workflows, AI will become more than just a helpful tool. It will act as a virtual coworker, autonomously managing complex, multistep workflows across various digital platforms.

Imagine a world where your AI doesn’t just suggest an ad campaign but launches it, monitors its performance, and adjusts it for optimal results—all while we focus on higher-value tasks. This is the future that LAMs promise.

What Does This Mean for us?

For businesses, the rise of LAMs and AI agents offers a massive opportunity to increase efficiency, reduce costs, and drive innovation. As these systems mature, companies that embrace LAM technology early will have a competitive edge, leveraging AI to automate routine tasks and make smarter, data-driven decisions.

For individuals, LAMs will redefine how we interact with technology. Instead of relying on AI to just answer questions, we’ll soon be working alongside AI agents that perform complex tasks on our behalf.