Slides: 0vphxNt4wyk
Source Video
[](https://www.youtube.com/watch?v=0vphxNt4wyk)
Relationship To World's Fair 2026
These slides are extracted from a public AI Engineer YouTube video connected to World's Fair 2026. Speaker-matched clips are supporting context unless later confirmed as exact session recordings; official livestream recordings are day-level/event-level source material.
Related Scheduled Sessions
- No individual scheduled session mapping has been assigned yet; treat this as an event livestream deck.
Extracted Slides
OCR text:
LAB & PLATINUM SPONSORS
Amazon AGI Lab ANTHROP\C Google DeepMind MINIMAX
OpenAl ZB ss akamar arize 2YS %Braintrust bright data
B Srowserbase %docker -NEO4) ORAcLeE PayPal ¢&
-y reducto (yy Sonar’ © togetherai (ff Unblocked €} WorkOS
OCR text:
*
ws
Sea ceed
| World's Fair
OCR text:
chad Vibe Checks Fail in Production
World's Fair |
Scie
Engineering the future of Al
OCR text:
cece Agents We Use vs. Agents We Build
World's Fair |
PRESENTED BY spy meee oy OT ea ten ae]
aw Microsoft re cg igi tatngto
TRACK 5 * JULY 1, 2026
Evals
OCR text:
CACYOATA SkillsWork?(SkillsBench1.1)
ted skills jump resolution rate from 33.9% = 50.5% (+16.6 pts)
Curecod hurranarxten sidis boost task resolutien by +16.6 polnts overal (33.9% - S0.sx).
0733
TRACKSOJULYO2026 Evals
OCR text:
Worid'sFair CEffiCact WatHinGs Self-Generated& Bloated Skills Fail
Gnlt tEhoth vs. Parottahce!
Fast, low boken overhesd, hgh pe Cornpsct (c 200 Ines)t +19.0
Optimal baerce of guidarct i. Stnderd (200--500 Pnes: +
Cnstruction grit begins to degrad Ceteled (500-1000 Knes)
Comprehenria (> tooo Ine
Self-generated skils hurt aocuracy -8.t to -f1.5 point loss Blost burre rth
TrackGoJuLyt,2026 Evals
OCR text:
ice Triggering Skills: Model vs. User
aceasta ls | ee
TRACK 5 * JULY 1, 2026
Evals
OCR text:
an Gemini Interactions API Skill
World's Fair
SO Ot Rn Saar TO
7 a a ~ Cer Bats SI A HE c Ee
oe | PODS tre pp tee gees - . 7 - pees
TRACK 5 ¢ JULY 1, 2026
Evals
OCR text:
AIEWORLD'SFAIR-2OMINUTETALX Don't Ship Skills # Run skill evals with ablotion test
WithoutEvals eval_runner run evals/dcploy_eval.yoml --with-vs-without-skills\
--runs=3
Building Reliable Agent Capabilities Ski11Lift:+35.2%（Readytoship!)
Philipp Schmid, Staff Engineer, Google DeepMind
Engineering the future of Al
OCR text:
ehak chai Oe ed eared eel SES Oe TSC ee
eo
| ee ee |
TRACK 5 * JULY 1, 2026
Evals
OCR text:
elas aheil Siglo cuNcagialetins |< hOmnTe 14 [o>] RTI@MMCt iT
World's Fair |
TRACK 5S * JULY 1, 2026
Evals
Slide-Derived Subjects To Review
Subject extraction uses video title, related session titles/descriptions, transcript context, and OCR text when available. OCR is best-effort and should be reviewed against the embedded slide images.