Uploaded July 2025 | Updated September 2026, 2 weeks ago
In this talk from Arize:Observe 2025, Ishan Singh (Sr. Generative AI Data Scientist, AWS) and Amanda Lester (Sr. GenAI GTM Specialist, AWS) share how to bridge the gap between technical capability and customer experience with AI agents. They present strategies for building trust with AI agents, introduce the open-source Strands Agents SDK, and reveal a new integration with Arize AI for agent observability and evaluation. Along the way, they offer practical insights on:
Designing agents with contextual relevance, adaptability, and memory
Using Strands Agents to go from prototype to production in hours
Tying agent outcomes directly to customer lifetime value
Leveraging feedback loops, AB testing, and continuous evaluation
Visualizing agent behavior with Arize Phoenix for maximum transparency
⏱️ Chapters
0:00 – Intro: The Experience Is the Only Feature That Matters
0:54 – Empathy Over Capability: What Makes Agents Great
2:00 – AWS Agent Stack: Q, Bedrock Agents, and Strands
3:40 – Why Strands Agents? SDK Design Philosophy
5:10 – Key Benefits: Open Source, Pre-Built Tools, Deployment Agnostic
6:30 – Integrating With AWS Services: Bedrock, S3, OpenSearch
8:00 – Observability Built-In: Traces, Logs, Guardrails
9:00 – New Integration: Arize + Bedrock + Strands Agents
9:40 – Building Trust: Contextual Relevance and Agent Memory
11:05 – Bring Back Data Science: Closing the Experience Gap
12:00 – Loops, Churn, and Connecting to Business Metrics
13:00 – Feedback as a Tool for Agents
13:40 – Continuous Evaluation & Alarming in Arize
14:30 – How To Get Started: 30 Lines of Code & Docs
15:10 – Closing: Documentation, Samples, and Getting Involved
🔗 Explore Strands Agents
strandsagents.com
👉 Want to stay up to date on trustworthy GenAI? Check out upcoming events at:
arize.com/community
In this talk from Arize:Observe 2025, Ishan Singh (Sr. Generative AI Data Scientist, AWS) and Amanda Lester (Sr. GenAI GTM Specialist, AWS) share how to bridge the gap between technical capability and customer experience with AI agents. They present strategies for building trust with AI agents, introduce the open-source Strands Agents SDK, and reveal a new integration with Arize AI for agent observability and evaluation. Along the way, they offer practical insights on:
Designing agents with contextual relevance, adaptability, and memory
Using Strands Agents to go from prototype to production in hours
Tying agent outcomes directly to customer lifetime value
Leveraging feedback loops, AB testing, and continuous evaluation
Visualizing agent behavior with Arize Phoenix for maximum transparency
⏱️ Chapters
0:00 – Intro: The Experience Is the Only Feature That Matters
0:54 – Empathy Over Capability: What Makes Agents Great
2:00 – AWS Agent Stack: Q, Bedrock Agents, and Strands
3:40 – Why Strands Agents? SDK Design Philosophy
5:10 – Key Benefits: Open Source, Pre-Built Tools, Deployment Agnostic
6:30 – Integrating With AWS Services: Bedrock, S3, OpenSearch
8:00 – Observability Built-In: Traces, Logs, Guardrails
9:00 – New Integration: Arize + Bedrock + Strands Agents
9:40 – Building Trust: Contextual Relevance and Agent Memory
11:05 – Bring Back Data Science: Closing the Experience Gap
12:00 – Loops, Churn, and Connecting to Business Metrics
13:00 – Feedback as a Tool for Agents
13:40 – Continuous Evaluation & Alarming in Arize
14:30 – How To Get Started: 30 Lines of Code & Docs
15:10 – Closing: Documentation, Samples, and Getting Involved
🔗 Explore Strands Agents
strandsagents.com
👉 Want to stay up to date on trustworthy GenAI? Check out upcoming events at:
arize.com/community
