WeAreDevelopers Logo WeAreDevelopers
Sign in
For You LIVE
World Congress 2026 Europe - Virtual Stage • Jul 2, 2026 • Session details

From minutes to seconds: Lessons learned optimizing multi-turn agentic workflows

Douglas Reiser

AI workflow latency doesn't have to bottleneck your scaling. Discover architectural tweaks, from token masking to deterministic fallbacks, that slash processing time from minutes to seconds.

Pause
Mute Enter Fullscreen
#1 about 6 min

Introducing the multi-turn CRM agent scenario

An overview of a CRM AI workflow designed to extract and link entities from conversational text.

#2 about 4 min

Analyzing the data staging orchestration pipeline

How the orchestrator separates intent between sub-agents and deterministic workflow boundaries.

#3 about 2 min

Measuring latency and token usage via tracing

Tracking step-by-step LLM inputs and costs to establish baseline performance metrics.

#4 about 3 min

Calculating latency through time to first token

How model initialization times and token generation speeds dictate overall system latency.

#5 about 9 min

Reducing output tokens by referencing text

Using indexed markers to prevent large language models from unnecessarily repeating inputs.

#6 about 2 min

Trimming context schemas to control reasoning

How converting dense JSON inputs to plain text alters hidden model reasoning layers.

#7 about 3 min

Masking redundant data tokens with placeholders

Substituting lengthy IDs and URLs with reference keys to minimize context bloat.

#8 about 3 min

Maintaining prompt structures for prefix caching

Ordering dynamic variables last to maximize cache hit rates across multiple requests.

#9 about 4 min

Decoupling workflow states from conversational history

Managing function parameters directly from execution states rather than orchestrator transcripts.

#10 about 4 min

Replacing marginal AI calls with deterministic logic

Limiting language models to unpredictable edge cases while handling standardized steps in code.

#11 about 4 min

Toggling explainability features to lower inference footprint

Stripping debug reasoning tokens from production prompts to directly decrease generation delays.

#12 about 2 min

Benchmarking quality degradation on smaller models

Scaling inference engines down and accepting minor feature lapses for substantial budget relief.

#13 about 3 min

Reviewing cumulative token and latency optimizations

Comparing the aggregate baseline times against the optimized branch configurations.

Matching moments

4:26 min

Optimizing AI infrastructure costs by maximizing token caching

Liran Hason Liran Hason · World Congress 2026 Europe

1:08 min

Rebuilding complex agent architectures to optimize token utilization

April Yoho April Yoho · World Congress 2026 Europe

2:01 min

Transitioning toward AI-first coding and managing token costs

Clemens Wasner Clemens Wasner +4 · World Congress 2026 Europe

3:21 min

Optimizing AI token consumption and running local language models

Chris Heilmann +2 · LIVE

2:27 min

Running agents efficiently at enterprise scale

David Soria Parra David Soria Parra +3 · World Congress 2026 Europe

3:06 min

Addressing the gap between coding assistants and complex workflows

Julia Kordick Julia Kordick · World Congress 2026 Europe

More from World Congress 2026 Europe - Virtual Stage

One Does Not Simply Make Architecture Decisions Alone thumbnail
36:15
about 2 months ago

One Does Not Simply Make Architecture Decisions Alone

Vladas Diržys

From Black Box to Glass Box : Bedrock AgentCore Observability thumbnail
21:15
2 months ago

From Black Box to Glass Box : Bedrock AgentCore Observability

Yasemin Aktürk

Leading in Rhythm: How Women Make Better Decisions with Biological Intelligence thumbnail
30:41
2 months ago

Leading in Rhythm: How Women Make Better Decisions with Biological Intelligence

Julia Grosse

Better Together: Leveraging Your Observability Tools as a SIEM thumbnail
39:06
2 months ago

Better Together: Leveraging Your Observability Tools as a SIEM

Mathias Palmersheim

ChatGPT Apps: From Chat to Checkout thumbnail
30:29
2 months ago

ChatGPT Apps: From Chat to Checkout

Aaron Koivunen

Interior Decorating (TypeScript Decorators) thumbnail
29:22
2 months ago

Interior Decorating (TypeScript Decorators)

Michael Dowden

Upcoming sessions on this topic

Open session

World Congress 2026 North America

September 24, 2026 · 14:10–14:40

Stage 1

Anatomy of an AI Request: Where Latency and Cost Are Really Born

Dan Fu

VP of Kernels at Together AI

Dan Fu
Open session

World Congress 2026 North America

September 25, 2026 · 09:00–09:30

Stage 7

AI Agents are Only as Smart as their Context: Building a Real-Time Context Engine at Intuit

Bharat Patel

Lead Software Engineer at Intuit

Bharat Patel
Open session

World Congress 2026 North America

September 24, 2026 · 17:30–18:00

Stage 6

No Single Model to Rule Them All: Building Resilient AI Agents Across Open & Closed LLMs

Emmanuel Acheampong

Senior Manager Developer Relations at Crusoe AI

Emmanuel Acheampong
Open session

World Congress 2026 North America

September 23, 2026 · 10:45–12:45

Stage 10

Agents That Own Their Inference: Building Production AI Agents on Dedicated GPUs

Duan Lightfoot

Sr. AI Engineer, Akamai

Duan Lightfoot
Open session

World Congress 2026 North America

September 24, 2026 · 14:50–15:20

Stage 7

When Agents Became Users: Rearchitecting Identity and Permissions for AI at Scale

Yoav Gal

Product Lead

Yoav Gal
Open session

World Congress 2026 North America

September 24, 2026 · 11:40–12:10

Stage 4

Taming Rogue Agents: Observability-Driven Evaluation for Production Reliability

Anagha Rumade, Anjana Umapathy, Apoorva Jaiswal

Anagha Rumade
Anjana Umapathy
Apoorva Jaiswal

Related videos

Rethinking Workflows in the Agentic Era thumbnail
27:07
World Congress 2025 • about 1 year ago

Rethinking Workflows in the Agentic Era

Eric Jadi, Rene Pajta

3 Ways to Rebuild the Data Stack for Agents thumbnail
28:02
World Congress 2026 Europe • about 2 months ago

3 Ways to Rebuild the Data Stack for Agents

Jordan Tigani

Ship Smarter Agents, Not Bigger Prompts thumbnail
28:51
World Congress 2026 Europe • about 2 months ago

Ship Smarter Agents, Not Bigger Prompts

April Yoho

Context is all you need thumbnail
24:18
World Congress 2026 Europe • about 2 months ago

Context is all you need

Raphael De Lio, Samuel Agbede

What 500+ Production Environments Taught Us About Shipping AI Agents thumbnail
23:05
World Congress 2026 Europe • about 2 months ago

What 500+ Production Environments Taught Us About Shipping AI Agents

Liran Hason

Why Token Reduction Isn’t Cost Reduction - Dave Anderson & Sarel Weinberger, PhD. thumbnail
41:09
Coffee With Developers • 20 days ago

Why Token Reduction Isn’t Cost Reduction - Dave Anderson & Sarel Weinberger, PhD.

Dave Anderson, Sarel Weinberger, Phd

Related articles

From learning to earning

Jobs that call for the skills explored in this talk.