Explaining Large Action Models: How Do They Work?

Bringing together large action models (LAMs) and large language models (LLMs) will change how businesses use AI.

This mix will help artificial intelligence understand and act more easily, predicted to improve how businesses work, enhance the customer experience, and help companies develop new products and services.

While LLMs are trained to understand words and phrases and create original, grammatically correct text, LAMs are advanced AI models that understand language and can ‘think’ through tasks to get things done.

They can handle different kinds of information as multimodal models, such as pictures, videos, and sounds, making them work more like how humans use digital content.

As Nicholas Rioux, CTO, Labviva, an AI-powered digital purchasing platform, puts it:

“When used in unison, LAMs and LLMs can transform the way we interface with technology.”

All technology uses abstraction, which means showing only important details to make things simpler and faster, according to Rioux. Writing direct code for everything is too slow for big projects. LAMs and LLMs create a new way to simplify how humans interact with systems.

“Ultimately, these models will become the new interface for humans and the engineered world,” he says.

Techopedia explores the combination of LLMs and LAMs and what we expect might happen next.

How LLMs and LAMs Work Together to Help Businesses

When LLMs and LAMs work together, they combine decision-making with taking action, says Amar Ramudhin, professor and program lead of information systems engineering and management at Harrisburg University of Science and Technology.

This partnership can change how businesses not only plan but also respond to data and insights instantly.

For example, in healthcare, while LLMs can analyze a patient’s medical history, lab reports, and the latest medical research to recommend treatment plans or predict potential health risks, LAMs can help by automating real-time monitoring devices, such as insulin pumps, heart monitors, and even robots that can assist during surgeries or administer medication with precision.

Ramudhin said:

“In a hospital setting, LLMs could help diagnose diseases by analyzing patient records and medical research papers to recommend potential treatment paths.

“Simultaneously, LAMs could automatically adjust a patient’s treatment by controlling robotic medical devices that deliver precise doses of medication, monitor vitals, and alert doctors in critical situations.”

LLMs and LAMs working together form a complete AI system that does more than just understand the world — it acts on it, says Cliff Jurkiewicz, VP of global strategy at Phenom, an HR technology company.

“This synergy is what businesses need to realize the full potential of AI,” he says.

“LLMs provide the language comprehension and creativity, but LAMs are the muscle that transforms that understanding into concrete, measurable results.”

When you combine both, it goes beyond just basic automation; it leads to a complete change in how operations work, according to Jurkiewicz. AI agents that use LLMs and LAMs can change entire workflows by automating processes from beginning to end, answering customer questions, and performing complicated tasks.

“Businesses that don’t adopt this level of AI integration won’t just fall behind — they’ll become irrelevant,” he adds.

“To gain the full ROI on AI, companies must invest in agents that can both think and act.”

How the Integration of LLMs and LAMs Will Reshape Industries

Combining LLMs and LAMs will lead to big changes in many industries by automating complicated tasks that used to need human help.

Ramudhin says that businesses can expect:

• Cost reductions: By automating tasks, reducing errors, and using resources more efficiently.
• Improved decision making: LLMs offer information and LAMs take actions based on that information, leading to faster and more data-driven decisions.
• New business models: By bringing together understanding and action, companies can create completely new services, such as automatic customer service systems, self-running warehouses, or AI-powered product design.

Combining AI that understands language with AI that can take action will help companies create more flexible and powerful tools, changing how many industries work, says Vikas Shetty, head of product at bot management and account security firm Arkose Labs.

“LLMs are great at creating content, offering real-time help and analyzing information, whereas LAMs handle complex processes, manage workflows and execute multi-step tasks.

“When harnessed together a synergy is created – LLMs will handle content creation and data analysis, while LAMs will take care of automating and managing tasks, cutting down on manual work and reducing error.”

How Companies Can Take Advantage of Using LLMs and LAMs Together

Ramudhin says that organizations can start using LLMs and LAMs together by:

• Identifying business areas for automation: Businesses should identify tasks that involve a lot of data analysis or need the same manual actions repeated often. These tasks are great opportunities for using LLMs and LAMs.
• Building AI ecosystems: Organizations should invest in combining LLM and LAM platforms that work smoothly together across different departments, such as customer service and supply chain management.
• Experimenting with AI in pilot programs: Begin by trying out the integration in one or two processes, such as automating customer service with LLMs or managing inventory with LAMs.
• Investing in talent and training: Training employees in how to use AI systems and ensuring teams are prepared to work with these tools will help make the transition easier.

Case Study: How Google Duplex Provides Both Virtual Assistants and Actions

Using LLMs, Google Duplex can hold natural conversations to help book appointments, order food, or make reservations for users by understanding subtle language differences.

Adding LAMs to the mix, Duplex integrates with scheduling systems, food delivery services, and reservation systems to take action after understanding what the user wants

The result? LLMs handle the conversation with restaurants or businesses, while LAMs complete the booking or transaction autonomously.

According to Ramudhin, this integration will impact “simplified customer experiences, saving time, and increasing convenience in daily activities”.

Relying on LLMs Only

Currently, most practical generative AI systems rely on LLMs to handle text for many different applications, says Flavio Villanustre, global chief information security officer at data and analytics company LexisNexis Risk Solutions.

LAMs are a new area of study, and while the research shows great promise with the possibility of many new uses, designing and training them is much harder and more expensive, he says.

“I don’t think many organizations are currently using LAMs in production systems today, but this could change soon, once LAMs become available and more cost effective,” he adds.

However, early adoption is key, says Jurkiewicz.

“The companies that win will be the ones that experiment now, learn quickly and iterate as these technologies evolve. Start by identifying areas where AI can bring immediate impact — whether it’s automating workflows, enhancing customer interactions or streamlining operations.”

“The mistake would be to wait for a perfect solution,” Jurkiewicz says. “The perfect solution is the one that’s constantly evolving, learning and improving as it’s used.”

The Bottom Line

By using the capabilities of LLMs and LAMs together, businesses can build smart systems that work on their own, Ramudhin says. These systems can understand things well and take effective actions, changing the way companies operate.

Rogers Jeffrey Leo John, co-founder and chief technology officer at DataChat, a no-code, generative AI platform for instant analytics, agrees.

Combining LAMs and LLMs is set to change how businesses work, greatly enhancing customer experiences and leading to new and innovative products and services, he says.

“Industries can leverage capabilities of the LLMs and LAMs to create personalized experiences,” Leo John says. “The key to this transformation is harnessing AI’s dual strengths: comprehending human language through LLMs and orchestrating actions with LAMs.”

Read the original article here.