# GitHub Next & Tessl on the Self-Merging Repo

## Executive summary

The discussion outlines the evolution of software development from traditional CI/CD to a new paradigm: Continuous AI. Speakers presented models where automated agents handle code improvements, testing, and merging (Paul Stack). Key shifts include viewing continuous improvement as a system-level problem rather than an individual productivity issue (Don Syme), prioritizing fixing the build system over fixing the code itself (Patrick Debois), and leveraging advanced AI tools for knowledge retrieval and proactive information gathering (Robert Overweg).

## Key takeaways

- Continuous AI is the Third Pillar: The development process requires three pillars: Continuous Integration (CI), Continuous Deployment (CD), and continuous AI, which focuses on automated code improvement in the repository.
- Agent-Driven Merging Process: Advanced pipelines allow agents to open a pull request, pass multiple reviews/gates, push changes, and auto-merge upon successful completion. The UAT (User Acceptance Testing) gate remains critical for preventing regressions before end-user release.
- Focus on System Improvement: The primary mistake is fixing the code when an agent fails; the correct approach is improving the system that produced the faulty code. This shifts focus from 'fix the code' to 'fix the system.'
- Knowledge Retrieval and Briefing: AI agents can transform company knowledge into a searchable resource, allowing users to query complex information in plain language or receive daily briefings rather than managing a backlog.

## Technical details

- Continuous AI Paradigm: The concept proposes that continuous code improvement should be treated as a third pillar alongside CI and CD, moving beyond focusing solely on individual developer productivity.
- Automated PR Merging & Gates: A fully automated process involves agents opening the pull request, passing multiple review gates, pushing changes, and auto-merging. The UAT gate is highlighted as the final safety check that blocks releases until regressions are fixed.
- Metrics of Efficiency: In a 30-day period, five people were able to open 295 issues and ship 217 features for approximately $3,000 per month. The median time from triage through shipping was reported as 1.6 hours.
- AI Tools & Costs: The team utilizes Claude Code Max Pro (costing $200 per person) and spends an estimated $1,500 to $2,000 monthly on the CI review process.
- Systemic Improvement Mindset: The goal is to improve the build system itself (the 'thing that builds the thing') rather than manually correcting code errors when an agent fails. Observability and reusable specs are key enablers.

## Practical implications

- Organizations should shift their focus from individual developer output to improving the automated systems that build and test the code.
- Implementing robust UAT gates is crucial, as merging a PR does not guarantee readiness for end-users.
- Adopting AI agents for proactive knowledge retrieval (searching for ideas/context) can significantly reduce time spent searching through files or backlogs.

## Topics

Continuous Integration, Continuous Deployment, AI Agents, Software Development Lifecycle, DevOps, AI DevCon NYC 2026, tessl.co

Source: https://www.youtube.com/watch?v=9GArVPhHGkQ