Skip to main content

Startup Summit 2026: Sharing Technical Presentations

Juan F. Voltolini shares his first public speaking experience using a teammate's material at Startup Summit 2026, highlighting the value of hands-on demo replication.

AI-written
Inewgen
18 Sep 2026Source: Dev.to3 min read (0 views)
Share
Startup Summit 2026: Sharing Technical Presentations

Stock photo for illustration only, not from the actual event

Font size
  • Juan F. Voltolini delivered his first public talk at the Dati booth during Startup Summit 2026 in Florianópolis.
  • The presentation titled "Preço por token não é preço" was originally created by Bruno Vilardi Bueno of Dati Labs.
  • True knowledge transfer requires running demos firsthand rather than simply memorizing presentation slides.
  • Hands-on testing revealed new token consumption metrics between different model versions.

In late August, Juan F. Voltolini stepped onto a public stage for the very first time at the Dati booth during Startup Summit 2026 in Florianópolis, delivering two sessions across Wednesday and Thursday with a large screen showing live Amazon Bedrock operations while the audience listened via headsets.

The presentation topic, "Preço por token não é preço" (Price per token is not the price), was created by Bruno Vilardi Bueno, his colleague at Dati Labs, who first introduced it during a Dati webinar in June. The package included the slide deck, two live demos, a script, and an appendix addressing recurring questions.

Brazil technology seminar presentation screen

Stock photo for illustration only, not from the actual event

From an organizational perspective, any technical asset that relies solely on one specific individual creates a bottleneck, regardless of how polished the presentation slides might be. If the original author is unavailable, the content cannot be delivered. Bringing Bruno's talk to the summit served as Dati Labs' initial test to overcome this limitation.

Technical knowledge sharing within tech organizations frequently faces key-person dependencies. Overcoming this bottleneck requires more than documentation; it demands a process of internalizing information where the recipient reconstructs the knowledge into their own repertoire, ensuring sustainable organizational learning.

Never miss the latest news?

Subscribe to get news summaries by email - not often enough to be annoying.

โฆษณา

The inclusion of an appendix with prepared answers regarding measurement timelines, provisioning costs, and model switching gave Voltolini the confidence to handle live Q&A sessions during both rounds.

64%Token variance across different models for identical input
25%Token increase when updating models within the same family

While thoroughly studying the material and running the demos in his own account to understand the underlying numbers, Voltolini uncovered new data absent from the original deck. Sending an 11-word Portuguese phrase across eight models on Amazon Bedrock resulted in a 64% difference in token count, while updating to a newer Anthropic model family increased the token count by 25% despite sharing the same pricing table.

"Preço por token não é preço"

Bruno Vilardi Bueno

This newly discovered metric became the highlight that engaged the audience during both sessions, proving that executing the demos firsthand yields insights that simple memorization misses.

Source: Dev.to

Comments

Leave a Comment
0/2000

Found something wrong in this article? Report an issue with this article