Slides: Context Is the New Code — Patrick Debois, Tessl
Source Video
Context Is the New Code — Patrick Debois, Tessl
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:
€3 Braintrust €} WorkOS OpenAl
OCR text:
Za hop oa Context is the
f
ites
¢ whe 0 8 wees
OCR text:
ContextisthenewCode
Prompt→Code
Code→Skill
Prompt
Code
deffetch_user（uid,retrles3)1
for1inrangelretries)i
Addeorhandingandretry
tryi
logic wltha timeout.
r=requests.geti
AIE
f/users/(uid)",tineout=5)
r.raise_for_statust)
returm r.json()
packagedas
Code
Ski
det fetcseruid,retries3):
for1Inrange(retries)
AP1 Fetch Best Practices
tryi
Aways set arequest meout
rreguests.gett
-Retry on transient errors
(5)/s/
Re-raise on last attempt
r.ralse_for_statusl)
returnc.json[)
EvaLs
except Exceptien as e！
meout paramsset
f1-retries-1:raise
√retesz3
Prompt → Code (left) ·Code → Skil (right)
AlEngineer
Engineering the future of Al
AEnginoer
OCR text:
to think in parallels
AIE
2009-WhatifOpswaslikeDev
2025-WhatifContextwaslikeCode
AIEngineer
Patrick Debois-ProductDevrel Tessl
AIEngineer
AEngineer
EUROPE
OCR text:
Context Development Lifecycle
Generate Distribute
(XD)
Evaluate Observe
|
a
Engineering the future of Al
wmPuses
OCR text:
Prompts - Humans as context engine
Welcome back Patrick!
ask Claude to Create.
Penn 3
* of
aco
* af
ed
Tell re when my talk 15 at Al engineer = https: //wew.at.enqinecr/europe
Fetchihtips://we..ar.engineer/europe }
Received 447KB (2006 OK)
e@ Your talk “Context Is the New Code" 15 on April 10 (Day 3) at Ll:15am in the
Westminster track.
| ag |
ee]
r 2 °
a] Engineering the future of Al
we Pics
OCR text:
Rules, Instructions - Agent.md
AGENTS.md
Panel a
* Bd
. a ==>
* 5 d
os
| Lapa |
'
| AlEngineer |
Bun AM a
OCR text:
Context Connectors
rales a Pee Reis] 7 i _ i a °
“ee a as
aw
.
vee
Engineering the future of Al
OCR text:
Today's Agenda
Pane a ;
01 ¥ Generate — create & curate context
naa
5g *
nid 02 ® Evaluate — test & measure context quality
03 *® Distribute — package & share context
04 © Observe — monitor & improve in production
| Fa ed
[ aeal
. * ;
=] Al Engineer
Puss EUROPE
OCR text:
Skill syntaxcheck~Grammarly
AIE
end best pacisices) and sher
ctges
https://docs.tesslio/evaluate/evaluating-skills
AlEngineer
Engineering the future of Al
AlEngineer
OCR text:
Evals(LLM asJudge)
CLAUDE.md-ContextRules
ArI Stasdatds
Erery A: eodpoint muat uae the prefix/e
AIE
Prompt
esneesoluodpumaueoppy
eapp.post(/eesome/user)defsave_user()1
oeserate
Eval-LLMasJodge
Checki Does thepe
for saving a user
/PASS
CLAUDE.md rule - Prompt - LLM-as-Judge eval
AIEngineer
Engineering the future of Al
AlEngineer
OCR text:
Pee G
e - Non-déterministic ¢ untestab! Nk error budgets
acd
* x « © Run 5+ trials, force binary decisions, not fuzzy scores
bd * bd
Care atest OID SS fal full su:te on Cl > scheduled drift checks
¢ @ Mine productidn Taillires — they re your Loi data
a
° @ Every context change reruns the suite - - ndtkcepticns —_ f
. Vendor metncs le — define your own
Led
[ a 5 5
Cm] Engineering the future of Al
Piss
OCR text:
Public vs Private Registries
a: @
BRKILLS The package manager for
rales ; : agent skills and context
bd bd
nares
rs * pn
wy *
| lachal
a Google DeepMind
Piss
OCR text:
DependencyManagement
Tessl Docs
Pvess Kt
Q
WhyAPM
Alcodid
That drectorylslinked to tessl.JsonThistleprovides documentationfor the FastAPI
ody ryoper stsumyngptbenrrrdublee
nue
nomaniest for
AIE
tool wenyouneedinfomationabout FastPIusagefeturesand bestpractices
APMfisDec yorprjc'santc dndecesonceyand eey
Inidethetledectyyo/fdamaietflecaledcntis
dddopdep aut
-spuosudasdepngryeaodaofsuopoyadoap
meadataabouthelencldingthpacae
ape.ya1 - ihp vith your groect
name:your-project
"vorsiono16.o.
"same:tessLpypl-faatap1*,
dependencies:
"docs:docs/Lndex.ad'.
sitls froeay reositooy
"describes:pkg:pyp1/fastapi0.116.1*,
FPlugins
threpics/skilts/skitls/frentene-design
"summary':"FastAPI franework,highperform
fast
-gith/esome-copilot/glagins/context-englneering
Specificagentgriaitiesfrosey repoiitory
1fllatcsr
-eicreseft/aoe-sanole-gacage
https:/docs.tessl.io/use/make-your-agents-smarter-with-
gitclme-erg/repo-cd-repe
documentation
APM -https://github.com/microsoft/apm
AIEngineer
AIEngineer
AlEngineer
EUROPE
OCR text:
Context security scan
ann)
ae
* a
a cee
a
ca | AlEngineer |
a—Disrs EUROPE
OCR text:
4 Learn fromA
Aqeat Tract
| AlEngineer
ist ead
Le
| AlEngineer
fUROFE 4
) a 7 a
es in oe
OCR text:
S Entire | signin |
air Every commit tells a story.
i bs] 4 Now you can read it.
oa
:
t 5 .
Engineering the future of Al
Pass
OCR text:
Context from production code
Calera]
a e Pa * Understand code behavior with
* function level data
[serge]
|
k a
Engineering the future of Al
we Pais
OCR text:
Agent sandboxing
@ BEERNS LASERS 5 SEARS STENTS
Pen irangtié
va * L7 Observabilty & Auait TY Oate Exfittranon abe egt
LG Action Governance T2 Supply Chain $ Cooperative
re ry .
rae LS Credential & Secret Mgmt 13 Oestructive Ops 2 Soltewrare-entorced
Ld 3 Kernel-enforced
La Network Goundary T4 Lateral Movement 4 Structurad
L3 Filesystem Boundary T& Parsistence SESE MG
L2 Resource Limits 6 Privilege Escalation 1 Binary (onjott)
Li Compute Isolation T? Denial of Service Zane bust
1 Pet-resource policy
| Ea
eaort
cn er
= |
—
—_~
A
C=] Al Engineer
Piss EUROPE
OCR text:
Skills Development lifecycle
Oca as teal O strioute Onseve
* rs
nN
Ss a
| aaleplniptal
$3 Braintrust €) WorkOS OpenAl
a Piuss
OCR text:
EZ
Context Flywheel .g.
“ a
; a A 888 A cee
a * © a OSS \ re
* —~ BRBNY SS =
ae ty YS; > . Tees SET OF TEAMS Ws:
C&S) =
\ _~ x . ie ss pe. o/. 4
eee —
a
AlEngineer
ae) cela
OCR text:
, _ , eee
me 6=6Context FI
Al Engineer
3
ae rte
i ary mT Ss
ns, po ocue®
ee e
XQ %
-~ oe
[ AlEngineer
Wiis
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.