AI is everywhere—but most teams still treat it as an accessory, not as infrastructure.

Think about this; You use a database to persist information, you use an API gateway to expose structured data to other parts of the business. What function can you use an LLM for? You’re missing out on leveraging this technology!

In this talk, we’ll explore how LLMs can serve as dynamic building blocks—not just for ML specialists, but for everyday engineers. We'll move beyond the usual "AI for automation" narrative and show how LLMs can be treated like adaptable modules in your software architecture.

Drawing on real-world case studies from our work with clients, we’ll dive into patterns that signal LLMs might be the right tool—scenarios like:

• Systems with tons of heuristic rules
• Tasks that tolerate fuzziness but need scale
• Workflows with verifiable outcomes
• Data locked in PDFs or images

You’ll leave with a practical framework for spotting LLM-shaped gaps in your own stack—and a clearer idea of when AI isn’t just a solution, but part of the system.

AI is everywhere—but most teams still treat it as an accessory, not as infrastructure.

Think about this; You use a database to persist information, you use an API gateway to expose structured data to other parts of the business. What function can you use an LLM for? You’re missing out on leveraging this technology!

In this talk, we’ll explore how LLMs can serve as dynamic building blocks—not just for ML specialists, but for everyday engineers. We'll move beyond the usual "AI for automation" narrative and show how LLMs can be treated like adaptable modules in your software architecture.

Drawing on real-world case studies from our work with clients, we’ll dive into patterns that signal LLMs might be the right tool—scenarios like:

• Systems with tons of heuristic rules
• Tasks that tolerate fuzziness but need scale
• Workflows with verifiable outcomes
• Data locked in PDFs or images

You’ll leave with a practical framework for spotting LLM-shaped gaps in your own stack—and a clearer idea of when AI isn’t just a solution, but part of the system.

AI is everywhere—but most teams still treat it as an accessory, not as infrastructure.

Think about this; You use a database to persist information, you use an API gateway to expose structured data to other parts of the business. What function can you use an LLM for? You’re missing out on leveraging this technology!

In this talk, we’ll explore how LLMs can serve as dynamic building blocks—not just for ML specialists, but for everyday engineers. We'll move beyond the usual "AI for automation" narrative and show how LLMs can be treated like adaptable modules in your software architecture.

Drawing on real-world case studies from our work with clients, we’ll dive into patterns that signal LLMs might be the right tool—scenarios like:

• Systems with tons of heuristic rules
• Tasks that tolerate fuzziness but need scale
• Workflows with verifiable outcomes
• Data locked in PDFs or images

You’ll leave with a practical framework for spotting LLM-shaped gaps in your own stack—and a clearer idea of when AI isn’t just a solution, but part of the system.