Slides: The 100-Tool Agent Is a Trap - Sohail Shaikh & Ankush Rastogi, Prosodica
Source Video
The 100-Tool Agent Is a Trap - Sohail Shaikh & Ankush Rastogi, Prosodica
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:
Ankush Rastogi
The 100-Tool Agent
Is a Trap
Sohall Shaikh
Scaling with SemanticRouters and Just-In-Time Context
Al EngineerWorld'sFair2026
For Engineers BuildingLLM Agents
Ankush Rastogi
OCR text:
THEPRESENTERS
Ankush Rastogi
Ankush Rastogi
Sohail Shaikh
Sohal Shaikh
SeniorData Solutions Engineer.Prosodica LLC
DataScientist·Prosodica LLC
IEEESeniorMember
BuildingReal-WorldAl Systems
10+yearsacrossdataengineering,lsystemsproduction
9+yearsinAlLP,conversationalintelligence,AGpipeines
analytics,andenterprise LLMimplementation.
semanticsearch,andproductionLLMworkflows.
Ankush Rastogi
OCR text:
eae”
Pat
THE PROBLEM ret)
7 ed -
The Fat Agent Trap
The Natve Architecture
Cs
|] how
Token Bloat °. oa , ag
AccuracyCrash - + an rs
[eekia a telrestlola) ; fu REY y F
Context Crowding °. me ;
OCR text:
WHY IT FAILS it
Accuracy Collapses With Scale
ee Se
Tool Selection Accuracy (°%e} vs. Too! Pool Size ’
OCR text:
THETECHNIQUE
Just-ln-Time Context Injection
Ankush Rastogi
StaticLoading
JIT Injection
All100+schemaspre-loaded in theprompt
Toolsselectedatruntime,perquery
Everyrequestcarriesthefullpayload
Only3-5relevant schemasinjected
·Most tokenswastedonirrelevanttools
Contextwindow stayslean
Contextwindowconsumedbyschemas
VS
Moreroomforreasoningchains
Sohal Shaikh
Lessspaceforreasoningandoutput
Accuracy staysabove83%at any scale
Accuracy degradesasthelistgrows
Fast:smallerprompt,lessprocessing
Slow:model mustprocessagiant context
InspiredbyAnthropicMCPon-demand loading
Ankush Rastog
OCR text:
practel aS it
; — ae
Benchmark Results
Per eet
Accuracy (%) vs. Tool Count TTFT (ms) vs. Too! Count @ GPT4o
a ee
ee ‘ee
| bon
= ff
oe
od co
Pn ete
OCR text:
HOW TO BUILD IT it
3 Step Implementation Pattern
Build Tool Index 7 Route Each Query on Inject & Call LLM
a | ee ‘ia :
OCR text:
eaters a at)
. P — ‘ie
Implementation Checklist
Catalog Your Tools ., Build the Embedding Index
+ , |
; Implement the Router Integrate into the Agent Loop
7 Evaluate & Tune K 7 Monitor & Iterate
OCR text:
P= ae
INDUSTRY EVIDENCE it.
an p |
Teams Are Hitting the Same Tool-Scaling Wall
Cee cea Anthropic Engineering Blog eee SOK «Issue
- : S ? ma a
ee A |
MCP-Zcro (xfey/MCP-Zero) n8n Community Forum
Eo
OCR text:
KNOW THELIMITS
Trade-Offs&BestPractices
Ankush Rastogi
CONCERN
MITIGATION
Router maymiss a needed tool
Fallback:raiseK,or let theLLMrequestmore
Adds complexity:vector DB+tuning
Embedding search is ms-fast; the savings dominate
Sohail Shuikh
Raretoolsmayranklow
Logmisses;retrain oraddkeyword boosting
Kthresholdishard to calibrate
Start atK=5;tune on a dev eval set
Notworthitbelow~20 tools
For smalltoolsets,load statically;norouter
Ankush Rastogi
OCR text:
a!
REMEMBER THIS it,
oo eee —_
Key Takeaways
— Too! Overload Kills Accuracy ye ne -
- o #6 2 fe a 7 , ‘
“s Tokens = Money + Latency oe _ ] Mg ; a
aan Semantic Routing Saves the Day | _ , , oe
‘It's RAG, but for Tools ee a en
a Tele scTarl Mestre Otel talib rs - ,
OCR text:
GO DEEPER it
Resources & References .
aw] Kasia Ss a fetel mM <:] elel-} = API Docs
Thank you!
OCR text:
| (|
LET'S CONNECT ft oot
Thank You _
a
Ceo ee etl
ees ne
San fale Syn
Ankush Rastogi Sohail Shaikh
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.