In this article, we break down the key learnings from his session. The takeaway? Not just bigger models – but smaller, smarter ones, paired with more autonomous patterns of execution.

Small Language Models, Big Impact

Ahmed explained that while massive foundation models often get the spotlight, they’re rarely the best fit for enterprise tasks that require high precision, predictable outputs and fast execution at scale

Instead, Small Language Models (SLMs) – compact, fine-tuned models trained on specific domain needs – are enabling smarter automation across areas like invoice validation, reconciliation, and compliance workflows.

“We don’t always need the Ferrari of models. Most of what we’re doing is boring work – and that’s exactly where smaller models excel.”

By choosing models that are optimised for the task, not just the benchmark, Firemind is helping clients realise faster deployments, reduced costs, and greater control over how AI operates inside their business.

The Role of Agentic AI

The session also touched on the shift from static prompt-based systems to agentic patterns – AI systems that can think through goals, break tasks into steps, retry and adapt and operate with minimal human oversight

Ahmed framed this as AI that behaves more like a team member than a tool – constantly working toward resolution in a structured, autonomous way: “We as humans work in loops. These agents work in the same way – they keep going until the task is done.”

This evolution allows businesses to move from GenAI experiments to end-to-end process automation.

Practical, Scalable, Production-Ready

What made Ahmed’s session stand out was its practicality. Rather than pitching “AI magic,” he highlighted how Firemind designs systems that solve measurable problem fit into existing workflows, prioritise accuracy and reliability and avoid over engineering.

Whether it’s automating accounts payable or accelerating compliance checks, Firemind’s approach proves that less model can mean more impact – when it’s applied with purpose.

Watch the Full Session

Catch the full episode of AWS GenAI LIVE, featuring Firemind’s Ahmed Nuaman alongside AWS hosts Brian Terry and Wayne Saxe, as they explore how enterprises are turning GenAI into real-world value.