From stateless to stateful production-ready agents

[AdityaPainuli]

Talk title

How to make AI agents stateful

Short talk description

One of the biggest limitations in AI agents today is that they are stateless.
And this becomes a massive bottleneck as agents start performing heavy tasks, multi-step reasoning or long-running workflows.
My talk would focus on how to make simple design choices while building such agent to make sure even if the process goes down agent starts from where it left

Long talk description

How to solve the massive bottleneck of agents being stateless. especially in domains where they perform heavy tasks and multi-step reasoning and processing.
In case the system fails or the operation crashes, the complete process goes in vain.
We will focus on building a very simple, more deliberate, and with full in-house control:
A Redis-backed, fully durable, crash-safe stateful agent engine.

What format do you have in mind?

Talk (20-25 minutes + Q&A)

Talk outline / Agenda

• Introduction to the problem ( Agents stateless)
• Our simple redis-based arcihtecture to solve the problem
• Some advance redis concepts, such as locking and ownership in a distributed system
• Walkthrough of building a minimal AgentStore class that makes the flow persistent and resumable without any 3rd party tooling.

Key takeaways

• Understanding mental models of building a reliable agent
• Advance concepts of Redis in any distributed production system
• How real-world agentic system scales and think

What domain would you say your talk falls under?

Core Python

Duration (including Q&A)

30 minutes (20 minutes talk + 10 minutes Q&A)

Prerequisites and preparation

• Basic knowledge of LLM and Python.

Resources and references

I have the code in typescript either I can represent this or can rewrite it over in python but would prefer to keep the concept and implementation as same - https://github.com/AdityaPainuli/stateful-agents

Link to slides/demos (if available)

No response

Twitter/X handle (optional)

@aditya_painuli

LinkedIn profile (optional)

https://www.linkedin.com/in/aditya-painuli-422996218/

Profile picture URL (optional)

No response

Speaker bio

• Building low-latency AI agents with different startups and agencies for around 3 years now.
• Working in a European AI company that performs moderated AI interviews and thematic analysis on them
• On one side, building a real-time platform that provides 1:1 tutor sessions with animation and visuals - voiceprepai

Availability

24th January

Accessibility & special requirements

No response

Speaker checklist

• I have read and understood the PyDelhi guidelines for submitting proposals and giving talks
• I will make my talk accessible to all attendees and will proactively ask for any accommodations or special requirements I might need
• I agree to share slides, code snippets, and other materials used during the talk with the community
• I will follow PyDelhi's Code of Conduct and maintain a welcoming, inclusive environment throughout my participation
• I understand that PyDelhi meetups are community-centric events focused on learning, knowledge sharing, and networking, and I will respect this ethos by not using this platform for self-promotion or hiring pitches during my presentation, unless explicitly invited to do so by means of a sponsorship or similar arrangement
• If the talk is recorded by the PyDelhi team, I grant permission to release the video on PyDelhi's YouTube channel under the CC-BY-4.0 license, or a different license of my choosing if I am specifying it in my proposal or with the materials I share

Additional comments

No response