Slides: The Prompt Is Still a Punch Card - Ted Johnson, JoinIn AI
Source Video
The Prompt Is Still a Punch Card - Ted Johnson, JoinIn 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:
oe
| at sf |
rs
e |
| \
OCR text:
,
ean, oe ea
A Re i oo
ib RY | clr
PY N css: fo
J 'SQO@
wes — F
a @
OCR text:
:bytes
sFind Python files>18
Variables
Endpoint
timit100
What would you like to build?
sfind.-typef-name*py-size·k
554889E54883
itens-[]
https://api.example.com/v1/users
EC20488B7DF8
sCount Tooos in those files
Functions
Method
Write aPythonfunction
488B75F05DC3
$grep-RInTO0O1wc-1
def fetch（url,timeout-5):
GET
that parses a CSVfile,
resp·http.get（url,tineouttineout）
deduplicatesrowsby
return resp.json()
Auth
email (case-insensitive),
sSho top5by size
Conditionals
Bearer Token
keeps themostrecent
；x86-64
s1s-Uhs（find.-name'.py）
ifresp.status_code200:
record byupdated_at,
push
rbp
sort-kS-hr|head-n5
Token
and returns theresult
mov
rbp,rsp
-re-r--r-.
else:
18k Lib/parser.py
raise Error(resp.text)
asaJSONarray.
sub
rsp,8x26
-W-F--r--
-r-r----
12xapi/elient.py
mov
rdi,[rbp-0x8]
8.0Kutils/i0.py
Loops
(NOsr） son
Requirements:
rsi,[rbp-0x10]
-rw-r--r-.
000c
mov
6.1K tests/test_api.py
for Item in data:
Use csv.DictReader
-r--r-
ifiten["active"]:
rbp
.Stream large files
pop
sFetch API andpretty-print
itens.append（iten)
'actiw':trve,
"iit:100,
.Validate emails
ret
5curl-shttps://api.exanple.com/\
upe,:olo,
Sort by updated_at desc
users?linit-31j0.
Include a count summary
Add type hints and
（p.PT.）
Data (JSoN）Primitive
Timeout(s)
docstring
（"id:2.name²:Linus}.
user.
（"id":3.name²:Grace°）
"email:adaexanple.con
"role":'adnin"
Response Format
JSON
0x01
Opcodes
Shell
Primitives
Fields
Open Language
OCR text:
PROMPT
Tellmewhatmatters.
JUSTASTRAWINTOTHEOCEAN.
OCR text:
. FF oaern
i. | A
Protocol
OCR text:
| hh
———<$<$<————————
——— ae
= | ;
OCR text:
| |
sioer>1U\om iG) g(0 16] (01 an malci=1e MU lcm COm=1 a 0nl0) Kes
It can ask a follow-up. It can clarify mid-thought.
It can notice it's missing something and say so.
It should be human conversational
OCR text:
Batch Car:
1 LOOM ? PUNCH CARD Omens ast rs 4 PROMPT
SO,
/ °o
ean ce area Oe Cemeeee Cc PRIC OLE as cee ees ener ee aR
aera Sore aoe eee iaal® Pa cae
ex en eran STO ers rea: gee” PRESS 4 eared | coe. eee
NCR ene @-2 Ee 8-2 ak ok - ea Gaee cee °
OCR text:
Expressivegrowth
thegap
OW
Punchicards
Command ine
asnou+no
Touch
SWTT
esopgaobasd
Channel-how he bits physically move
Inleraction protocol -whoholdsthe floor
OCR text:
LI : | iv ‘
7 ‘| ry 7 ;
h an
OCR text:
ne © Frontier Voice
@rse
OCR text:
atl © Frontier Voice * ©
oO You v1ePM
When's the next 3
& Timberwolves game?
g Frontier Voice 2 12 1"
3 hep ruet Mt co dte Teta ater dummies Nat
ry ee at
fas! e € we gu D
ou UB
asy Ted centone 3
diy Frontest Voiced .
Sue Cr hee WEhans on pa fend?
3. Mes age Frente voce d +]
OCR text:
agGie) 4
» OpenAl is reportedly testing an unannounced bidirectional voice model called “GPT-Bidi-
Me
» Code references and early user tests show that the model can speak, hear, and listen
simultaneously, handling mid-sentence interruptions naturally.
= The unannounced model has already started rolling out to a select group of app users,
hinting at an official release window this week.
OCR text:
Enriching PersonaPlex’s output with non-verbal aspects creates an important qualitative
difference relative to systems without this dimension: PersonaPlex now recreates some of
the same cues humans use to read intent, emotions, or comprehension.
Examples
NVIDIA ADLR The following examples showcase PersonaPlex’s behavior across different scenarios. In
all audio files, you can hear the user speaking in the left channel and PersonaPlex in the
right channel (shown in green).
1 Assistant
1 I . i vt
Prompt: You are a wise and fnendly teacher. Answer questions of provide advice in a clear and
engaging way.
In this example from FullDuplexBenct’s interruption evaluation, PersonaPlex
demonstrates general knowledge, interrupterability, and natural turn taking.
2 Customer Service - Banking
Prompt: You work for First Neuron Bank which is a bank and your name is Sanni Virtanen
Information: The customer's transaction for $1,200 at Home Depot was declined. Verify
customer identity. The transaction was flagged due to an unusual location (transaction
attempted in Miami, FL; customer normally transacts in Seattle, WA)
OCR text:
Conversation is more than turn-taking
Whatreal-time conversational Al has to understand
1.PARTICIPANTS
2.FLOW
mm-hm
Turn-taking
Backchannels
Whoishere
Who isspeaking
Interruptions&repair
Timing
Whoislistening
Who isbeing
addressed
Natural conversation=
timing+context+
Overlap
repair+social understanding
3.MEANING
4.ACTION
Attention
Thread tracking
Decisions/
Whento speak
Grounding
Shared context
Beliefs/
commitments
vswait
intentions
OCR text:
Product requirements review uN
Bo a %
° a Pee Gulla)
ear Aa) arerze a)
It joins like a teammate — and starts by listening.
OCR text:
cca EE a ee eC ead a
£ Fo EA
, ms Presenting
Oana a) Jordan
ee at ee ea
Le a il
MT Re tne!
, Pulling up the source — so the room sees it too.
OCR text:
Product requirements review u
a Speaking
rn
3 o @)
3 A
_»
Oana on STE . ,
(er ees ee
. A decision — nota transcript.
OCR text:
Product requ:rements review a
oa
Bo Fo x
a as *
ferred Sai alt E ra)
lett San sate era eee a eee)
: It updates the spec — live.
OCR text:
y weg :
‘a en:
Loops, Prom:
useful patterns but st comstraming (oe nertact
BEYOND THE CONSTRAINT
en ae1@] <2.) 2? PROMPTS caer Ne] a) BES) _ ae
| ar
7 7 a mao y an
a | am n a
aa ,) conversation multimodal context-aware
- arn a
preted pad flay eae a ers Carentan Sie Caer eem St os —* l
SS Oar ary submit + repay Pee ee aiense or :
Si ace rghtaffordance timing silence
when useful
Design for what is possibic, not just what old interfaces .
OCR text:
a. see
a 4
“ie
seca » y
" @
OCR text:
<_< + |
Gap -* —
F Ne
iy
7 a,
ie
b = ‘
rf a :
CA@e sty
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.