Beyond AI Coding: Build an Autonomous Product Loop
See PM and developer agents plan, build, test and iterate on a real product
Claude Code, Codex, Cursor and other AI coding tools can dramatically accelerate development.
But they still require constant attention.
You define the task, provide context, review the output, request changes, test the result and decide what happens next.
The founder remains the bottleneck.
This practical workshop explores the next step: agentic development loops, where specialised AI agents work together as a small product team and continue moving the product forward with less human supervision.
What you’ll see live
Alex and Kate will demonstrate two AI agents working together on a real product.
The PM agent will:
Analyse the product, user signals and current priorities
Identify product and UX issues
Propose and prioritise improvements
Test user flows in the real product interface
Review the Developer agent’s work
Accept or reject completed tasks
The Developer agent will:
Implement changes
Perform technical checks
Run unit tests
Create pull requests
Record evidence of completed work
The agents communicate through GitHub Issues or Linear, retain context through shared memory and continue working through a full product-development loop:
Research → Prioritisation → Implementation → Testing → Review → Next iteration
The founder sets the direction and remains responsible for key decisions, without supervising every individual action.
How agent loops work
Before the live demo, Alex and Kate will give a short introduction to:
What an agent loop is
How the approach evolved from single-task AI assistants
How specialised agents collaborate and pass work between each other
How shared memory, testing and feedback connect separate tasks into a continuous development cycle
The workshop focuses on the practical architecture behind the system rather than high-level theory.
Meet the speakers
Alex
Alex is a Strategy & Ops Manager at Revolut with a technical and startup background.
He builds with AI daily and implements agentic and autonomous workflows in both personal projects and professional settings. He's co-built the agentic system demonstrated in this workshop.
Winner of a Google AI hackathon for developing agentic UI solutions
Kate
Kate is a Data Engineer at Dwelly, where she uses AI agents to automate software development, integrations and data migration processes.
She co-built the agentic system demonstrated in this workshop and is a Google AI hackathon winner for developing agentic UI solutions.
What you’ll take home
Every participant will receive access to a ready-to-use Git repository containing:
PM agent
Developer agent
Shared memory between the agents
Task handover logic
Automated testing
GitHub Issues / Linear integration
Decision history and test evidence
Setup and launch instructions
Local and remote deployment options
You’ll be able to clone the system and connect it to your own product.
Who it’s for
This workshop is designed for:
AI-native founders
Technical founders and cofounders
Hands-on early-stage CTOs
Solo technical founders shipping without a team
Founders already using Claude Code, Codex, Cursor or Copilot
This is not an introduction to AI coding tools.
It is for people who already use AI in development and want to move from supervising an AI assistant to orchestrating a more autonomous product-development system.
Format
A live, demo-led technical workshop with an optional hands-on component.
Bring a laptop with GitHub and access to Claude Code or Codex if you’d like to run the system alongside Alex and Kate.
You can use your own product or clone the prepared repository. A laptop is not required to benefit from the session.
Questions are welcome throughout the workshop.
Where the approach breaks
This is not a magic engineering team.
Alex and Kate will share where autonomous loops have failed in practice, including cases where an agent recursively created sub-agents without clear limits and rapidly increased model usage.
They’ll explain how explicit orchestration rules, tighter instructions and restrictions on sub-agent creation can prevent these runaway loops.
The workshop will also cover:
Iteration time and model costs
Incorrect decisions based on incomplete product data
Where human review and approval remain necessary
For MVPs, SaaS products and fast-moving early-stage companies, the approach is already practical enough to automate meaningful parts of the development cycle.
Enterprise applications require additional risk controls and guardrails, which are outside the scope of this workshop.
Tickets
In person in London - £75
In-person attendance is limited to 30-40 participants.
Your ticket includes:
The live workshop and demonstration
Optional hands-on participation
Direct Q&A with Alex and Kate
Access to the repository and setup materials
Drinks and networking with AI-native founders and technical leaders
Access to the workshop replay for 72 hours
The event will take place at a Central London venue. Full venue details will be shared with confirmed participants.
In-person applications
In-person attendance is limited and curated by the organisers.
The organisers reserve the right not to offer an in-person place where this is necessary to protect the relevance and value of the room.
We’ll review your application within two business days.
If we’re unable to confirm an in-person place, you’ll still have the option to attend online.
Online - £29
The online ticket includes:
Live access to the workshop and demonstration
Questions through the online chat
Access to the repository and setup materials
Networking and informal conversations will not be recorded.
Cancellations and transfers
In-person tickets are fully refundable until 7 days before the event.
After this point, tickets are non-refundable, but they may be transferred to another attendee subject to approval by the organisers.
Online tickets are non-refundable.
If the event is cancelled by the organisers, all ticket holders will receive a full refund.
If the event is rescheduled, ticket holders may transfer their ticket to the new date or request a full refund.
—-
By joining the event, you’re allowing us to capture and share some of the great moments! Photos and videos taken may be used later to showcase our community and for marketing purposes, helping us make future events even better.