0

Your cart

TOTAL excl.
TOTAL incl.
Pay
Go back

From the LLM Explosion to a Sustainable Platform: Surviving AI Growth Without Drowning in Costs

REXAutomatisationDevSecOpsInfrastructure

2026-12-09 | 12:00 AM - 12:25 AM

Register

Information

How can we continue to accelerate AI adoption when every new application increases costs, dependencies, and the risk surface?

In a context of rapid growth, we had to make decisions quickly, sometimes with imperfect implementations, often relying on instinct. The starting point was simple: connecting teams to the available models. Very quickly, this straightforward approach revealed its limitations: vendor dependency, firm and difficult-to-challenge budget commitments, scattered security controls, and complexity pushed onto each product team.

We will share our unfiltered experience of taking back control of the experience provided to users and teams. On the agenda: provider & router abstraction, request routing, gateways, multi-contract strategy, prompt and sensitive-data filtering, caching, trade-offs based on our company values, whitelisting constraints and geopolitical dependencies, followed by the gradual consolidation of the initial setup into IaC and CLI tools.

This experience report will highlight failures as much as successes. Our preference for a reversible and imperfect solution, our aversion to rigid budgets and architectures, and the industrialization around one central question: “What user experience and level of control over agentic coding do we actually want to own?”

Participants will leave with a decision-making framework to assess the appropriate level of abstraction, an example of an IaC configuration using OpenRouter, insights into vendor dependency risks and security controls, and guidance on when to industrialize. This is neither a product presentation nor a miracle architecture, but a story of concrete decisions, trade-offs, and consolidation in an environment that clearly shows no signs of slowing down.

Speakers