Smarter engineering choices  — the full picture of what your cloud really costs, and how to turn spend into runway.

The challenge

Your GenAI feature is in production, the AWS bill keeps growing, and someone has started asking whether it’s paying for itself. A 2026 industry survey of more than 1,700 business leaders found that only 6% of companies attribute at least 5% of their operating profit to AI. Those companies are twice as likely to have a defined process for measuring its impact.

As solutions architects, we have to keep up with models, prices and quotas that change every few months. Our delivery team benchmarks models on customers’ own data, and those results shape what we recommend. One pattern keeps coming up: the fastest, the cheapest and the most verbose model are often three different models, and a public leaderboard won’t tell you which one fits your workload.

Cloud spend is runway, and every pound spent on the wrong model is runway you don’t get back.

About the session

This is a practical session, not a product pitch. We’ll walk through the process we use with customers to determine what an AI feature actually returns: benchmarking models on your own data, routing each task to the right model, and continuously evaluating quality once it’s live.

Along the way, we’ll walk through a real example from a startup customer: where its costs were coming from, and what changed once they were measured.

You don’t need all of this on day one. You’ll leave with a clear place to start and a way to build on it.

What you’ll learn

Benchmark on your own data, so the model you pick is based on your prompts, your quotas and your costs.
Route by task, so each workload runs on the model that suits it, with fallbacks in place.
Evaluate continuously in production, so you catch quality drift before your customers do.
Report in numbers finance understands: a quality score and a cost per request.

Who it’s for: CTOs, founders and FinOps leads at startups running, or planning to run, GenAI on AWS.

Featured Speaker

Aleksandra Jovovic

Aleksandra Jovovic

Senior AWS Solutions Architect, GenAI Specialist, Automat-it

Aleksandra is a Senior AWS Solutions Architect at Automat-it, working with startups across regions. Her projects range from cloud migrations and container platforms to high-performance computing (HPC), GPU training infrastructure and AI governance, with a particular focus on generative AI and Amazon Bedrock.

 

Whatever the project, she helps teams make architecture decisions that hold up on cost, quality and scale, and that they can explain to the business.

 

In this session, Aleksandra will share:

  • Why model choice should come from your own data, not a leaderboard
  • How to route each task to the model that suits it, with fallbacks in place
  • How to keep measuring quality and cost once your feature is live
Boaz Ziniman

Boaz Ziniman

Host & Series Guide

Your host and series guide, Boaz Ziniman (ziniman.com), opens each session with the industry context and his own perspective on the topic — setting the stage for the “Cloud spend is runway” conversation.

Register Now

Seats are limited and the Q&A is where the real value lands. Don't let another month's cloud bill catch you off guard.

Save my spot — Register for the webinar

FAQ

Is it free?
Yes — the webinar is completely free to attend.

Who is it for?
Startup founders, CTOs, engineering leaders, and product managers who want to turn cloud spend into runway.

Will there be a recording?
Yes — all registrants receive the recording afterwards, even if you can't attend live.

Can I ask questions?
Absolutely — the session includes a live Q&A, and audience questions are the best part.

About Automat-it

We are an all-in AWS Premier Partner helping startups grow with AI, DevOps, and data on AWS. Exclusively focused on AWS and startups, we support you through the cloud journey, from implementation to 24/7 operational support - delivering solutions to 1000+ customers that reduce time to market, unlock growth, and optimize performance.


600+ AWS certifications | 180+ engineers | 1000+ customers | 9.8 CSAT