Slides: Building a Smarter AI Agent with Neural RAG - Will Bryk, Exa.ai
Source Video
Building a Smarter AI Agent with Neural RAG - Will Bryk, Exa.ai
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:
aWS
ee)
@®Graphite W Windsurf MongobB
Mdaily £3 augment code WorkOS
OCR text:
E Personal
QQ Searcr
OB Craig . .
Biden One API to get any information from the web
Si Bewei nae
Lb “Eee, ‘search ‘crawling Janewor ‘roeoarch
e anew ee
Res Vein tes a eee fe PES APE eR ee
a ee A RE COOK
SB toarpes Lorary \ geese
E ox . . .
Dens "
t
OCR text:
Nara ae eee
‘ aa s an
iL
| i a Microsoft Qc?
OCR text:
AIE
Microsoft
smol?
OCR text:
AIE
Microsoft
smol?
OCR text:
Traditional search engines were built
for humans
Humans want to type simple Google built a keyword based
keywords to click a couple links algorithm for humans
A
| ee)
OCR text:
The space of possible queries
Keyword
“stripe pricing” Pure LLM
oN “Explain this
©} _— concept to me”
3 Semantic
“People in SF
who know
assembly” —
super-complex
“Find me every
“————._ article that
; argues X and not
Queries no one Y from an author
ba oughtto _ —~ like 2”
a 7
a a Microsoft §omol®
_ —_~ ae :
OCR text:
EXPLORER
agent.py
github_agent.pyx
aiengineer
defget_people():
oerv
ogent.py
text-True,
github_agent.py
types"keyword",
report.md
num_resultsa1,
urlr.resutts[e].url if r.results else
out.append（（nane,url))
AIE
return out
defagent_mark（itens);
streanopenA.chat.completions.create(
nodel-"gpt-40-min1",
]asabessau
streanaTrue,
S6
encodin
PROBLEMS
OUTPUT
DEBUG CONSOLE
TERMINAL
COMMENTS
[Tsubasa Kato]（https://github.com/stingraze)
zh
[RobertoBayardol(https://github.com/roberto-bayardo)
[EdgarMe1j]（https://github.com/ejne1j)
Python
[DanlelCanpos]（https://github.com/danlel-e-campos)
[AuSH QuUATI]（https1//github.com/ankushagarva/ankush.cc)
[AnrAadallah]（https1//github.com/awadalLah)
[Antonio MalLsa]（https://github.com/amal（La）
Shashank
Ranaprased](https://github.con/shas
(scBiue
res
NG
villbrykgwills-MacBook-Pro-2alengineer
oo0ciadetau
Ln26,Col15Spaces:4UTF-8LF
()Python
Microsoft
smol
OCR text:
OS eprmm ommaner Btw tate re et eer
. FAL he dears oe ene ieret ieee Tg Cee NEI ENT A ttere ee SUee T font 3 @
BL ree od
Search
=.
Owens
Ca ne Deeg tee aren AL ‘
os = Meee ee ae Smee eyed
ae 4
oe feet depes .
Sq Set oe Nt tree Ae
gyeae aeeweg Sateen
‘ mt . Ptew sel Ave tee “
oon Cree One ergo i
tenes tee seca tee
fewer
Fe van taet
Crewing
om Meer Folens Nowe
em ©
Vareren rare rere ten toot inst
Ce
rosa I cans
a Microsoft § Gmc
Ps ri
' i a .
OCR text:
EXPLORER
agent.py
Aduerqra
angineer）agent.py>search_web
aengineer
oerv
agent.py
github_agent.py
repor.md
AIE
PROBLEMS
OUTPUT
DEBUG CONSOLE
TERMINAL
COMMENTS
Keyboardtaterrupt
zh
willbrykrills-MacBook-Pro-2alengineer
Python
willbrykglLs-MacBook-Pro-2alengineer
willbrykgwlLs-MacBook-Pro-2alengineer
Running agent
Personal Mebsites of Engineers in San francisco Interested in Information Retrleval
Searching...
Jonathan Koren
-Ena1tf1rstnanegjonathankoren.coe
-ebsite:[jonathankoren.com]（https1//jonathankoren.com/)
fes
NG
:fqdebog.
000cradeta
Ln15,Col 21Spaces4UTF-8LF
()Python3.13.0
Microsoft
smol?
OCR text:
EXPLORER
agent.py
github_agent.py x
UNTm
s
aengineer
exa=Exa（api_keywos.getenvEXA_API_KEY))
oerv
opena1-OpenAI(ap1_keyos.getenv("OPENAI_API_KEY"))
agent.py
deo
print(*GETTINGPROFILE NAMES")
res-exa.search_and_contents(
texteTrue,
AIE
type-"neural",
num_resultse10,
names1
for 1,hit in enumerate(res.results):
print（fEXTRACTING NAES（1))
name=openai.chat.completions.create(
model=gpt-40°,
messogesa
("role":"user",
fro the text.Only output the
"content:hit.title+"\n"+hit.text[:2e]),
othtng else.
PROBLEMS
OUTPUT
DEBUG CONSOLE
TERMINAL
COMMENTS
-tred-cli*: CLI utility for secure data storage.
zh
Exa-AI SearchEngine
Python
Funding*1Raised s17 mllion in Serles A to develop anAI-centric search engine.
on precision and better information retrieval.
s
-Will Bryk discusses the future of search ina sesslon highlighting technologicalinnovati
/atche9509qzbrQqu).
do
Key Takeays
-The shift froe traditional searchengines toAI-driven solutlons is seen as essential for
-Developeenton Gitubincludes coetributfons fron Bryk focused on supporting Cryptography
fes
NG
wiltbrykwills-MacBook-Pro-2alengineer
@o00cradeta
Ln26,Col15Spaces4UTF-8LF(）Python
Microsoft
smol
OCR text:
EXPLORER
agent.py
github_agent.py x
aengineer
oerv
model-"gpt-40",
agent.py
nessages=
dubeqrnp
{"role”:"user,"content"：hit.title+"\n”•hit.text[:2ee]),
othing else
).choices[e].message.content.strip()
Ifnane:
AIE
(aueu)puadde·soueu
print("nases",nases)
return List(dict.fromkeys(nanes))
defget_people():
out=[1
for name in get_nanes():
print（"GETTDNGGITHUB INFO")
r-exa.search_and_contents(
textaTrue.
PROBLEMS
OUTPUT
DEBUG CONSOLE
TERMINAL
COMVENTS
zs
names['Jonathan Koren',None','Doug TurmbuLl',“None',
'None','None','Aditya Varun Chadha',
Python
GETTDNG GITHUB INFO
GETTINGGITHUB INFO
GETTDNG GITHUB INFO
GETTING GITHUB INFO
[3onathan Koren]（https://github.con/}dkoren)
[DougTurnbulL]（https1//github.co/softaredoug)
None]（https1//github.com/none-None1)
Areyouusing a screen reader to operateVS Code?
[RobertoBayardo](https://github.com/rob
res
NG
wilLbrykgwilLs-MacBook-Pro-2alengineer
o00cimadea
Ln26,Col15Spaces:4UTF-8LF
(）Python
Microsoft
smol
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.