> Markdown version of [/videos/2136-moving-from-playing-with-ai-to-implementing-ai-lee-stott?t=2](https://www.wearedevelopers.com/videos/2136-moving-from-playing-with-ai-to-implementing-ai-lee-stott?t=2). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Moving from Playing with AI to Implementing AI - Lee Stott

Lee Stott warns that manual coding is dead, urging engineers to become multi-agent AI orchestrators. Master specification-led development, robust token governance, and localized model deployment.

- **Speakers:** Lee Stott
- **Event:** Coffee With Developers
- **Published:** July 29, 2026
- **Duration:** 18:43
- **URL:** https://www.wearedevelopers.com/videos/2136-moving-from-playing-with-ai-to-implementing-ai-lee-stott

## Summary

The transition from playing with AI to implementing it is shifting software engineering from manual coding to orchestrating agentic AI. Developer roles are fundamentally changing; rather than acting solely as individual contributors, engineers are becoming managers of multi-agent systems. This evolution requires a human-in-the-loop mindset focused on specification-led (or documentation-led) development, where crafting precise technical specifications and evaluating agent outputs takes precedence over writing raw syntax. Tools like GitHub Copilot, Microsoft Foundry, and Model Context Protocol (MCP) are equipping developers to focus heavily on architectural best practices, security, and the business justifications behind their code.

Organizations are increasingly recognizing the need for robust AI governance and guardrails. Tailoring fine-tuned models to internal company standards allows junior developers, who have often used AI coding assistants throughout their education, to be onboarded safely and effectively. At the same time, companies must manage token consumption and measure developers on the value and quality of outcomes rather than the sheer volume of code or products shipped. Benchmarking and evaluating models has become critical; utilizing tools like Model Router enables organizations to route basic tasks to smaller, cost-effective models while reserving frontier models for complex problems, avoiding unnecessary token bloat.

The ecosystem is also expanding to meet developers in their preferred environments, seamlessly bridging cloud and local constraints. Interfaces like the Copilot CLI and the new Copilot app provide unified access to LLMs, while localized AI is gaining traction through Windows ML and Foundry Local. By running highly optimized small-footprint models via ONNX runtime directly on developer hardware, teams can address compliance, security, and latency issues head-on, ensuring AI models are embedded closely within the application packages without excessive bloat.

**Keywords:** agentic ai orchestration, github copilot cli, microsoft foundry catalog, specification-led development, model context protocol mcp, ai token governance, localized llm deployment, windows machine learning, onnx runtime optimization, model router benchmarking, developer code observability, ai security compliance, multi-agent systems, documentation-led development, enterprise ai guardrails

## Chapters

1. **Moving from playing with artificial intelligence to business implementation** (00:02) — Developers are shifting from basic assisted coding toward orchestrating complex multi-agent setups.
1. **Evolving developer roles from individual contributors to agent managers** (02:04) — Human oversight remains essential as developers transition to evaluating automated code against architectural and compliance standards.
1. **Accelerating junior developers through organizational governance and fine-tuned models** (04:34) — Providing proper guardrails and specialized models allows new graduates to effectively evaluate code against internal company standards.
1. **Governing token usage and business objectives in software development** (06:40) — Aligning language model usage with key performance indicators ensures teams build features that deliver actual business value.
1. **Adopting documentation-led specification for optimal agent code generation** (08:11) — Investing in detailed markdown documentation and golden pathways prevents agents from generating unnecessary and costly code.
1. **Managing model fatigue through evaluation frameworks and model routers** (11:50) — Benchmarking systems dynamically select appropriate endpoints based on task complexity to optimize latency and operational costs.
1. **Interacting with code agents through command line and applications** (14:15) — Specialized command interfaces provide distraction-free environments for executing and evaluating specification-driven development tasks.
1. **Integrating localized artificial intelligence models within application packages** (16:30) — Hardware-optimized local models increase security and portability while avoiding massive file downloads for the end user.

## Related Moments

- [Developer role changes in the AI era](https://www.wearedevelopers.com/videos/100281-why-your-codebase-lies-to-ai) (from "Why your codebase lies to AI?")
- [Shifting developer workloads and realistic AI productivity gains](https://www.wearedevelopers.com/videos/1830-wearedevelopers-live-speculaitions) (from "WeAreDevelopers LIVE - SpeculAItions")
- [The transforming role of developers in the AI era](https://www.wearedevelopers.com/videos/100337-user-1st-technology-2nd-stop-building-ai-nobody-uses-start-delivering-real-business-outcomes) (from "User 1st! Technology 2nd! Stop building AI nobody uses - start delivering real business outcomes")
- [Transitioning software engineering teams to AI-native development workflows](https://www.wearedevelopers.com/videos/100087-ai-ready-what-enterprise-transformation-actually-takes) (from "AI-Ready? What Enterprise Transformation Actually Takes")
- [Adapting the software engineering role for AI collaboration](https://www.wearedevelopers.com/videos/100200-best-practices-for-ai-assisted-development-of-distributed-systems) (from "Best Practices for AI-Assisted Development of Distributed Systems")
- [Evolving developer roles into tech leads for AI agents](https://www.wearedevelopers.com/videos/1862-building-agents-securely-at-scale-alfonso-graziano) (from "Building Agents Securely at Scale - Alfonso Graziano")

## Related Articles

- [What is Agentic Programming and Why Should Developers Care?](https://www.wearedevelopers.com/magazine/625-what-is-agentic-programming-and-why-should-developers-care)
- [MLOps And AI Driven Development](https://www.wearedevelopers.com/magazine/82-mlops-and-ai-driven-development)
- [Navigating the AI Shift](https://www.wearedevelopers.com/magazine/629-navigating-the-ai-shift)
- [Transforming Software Development: The Role of AI and Developer Tools](https://www.wearedevelopers.com/magazine/527-transforming-software-development-the-role-of-ai-and-developer-tools)

## Related Jobs

- [Staff Software Engineer, Copilot Experiences](https://www.wearedevelopers.com/jobs/ext/164361-staff-software-engineer-copilot-experiences) at **GitHub**
- [Principal Software Engineer, Enterprise AI Platform](https://www.wearedevelopers.com/jobs/ext/1467292-principal-software-engineer-enterprise-ai-platform) at **GitHub**
- [AI Software Engineer (Germany)](https://www.wearedevelopers.com/jobs/48317-ai-software-engineer-germany) at **Sunhat**
- [Principal Product Manager, Agent Platform](https://www.wearedevelopers.com/jobs/ext/277541-principal-product-manager-agent-platform) at **GitHub**
- [Senior AI Agent Software Engineer (Go, Python) (m/f/x)](https://www.wearedevelopers.com/jobs/48277-senior-ai-agent-software-engineer-go-python-m-f-x) at **Dynatrace**
- [AI Operations Manager (all genders)](https://www.wearedevelopers.com/jobs/48263-ai-operations-manager-all-genders) at **envelio**