Transcript: ‘The AI Sandwich: Where Humans Excel in an AI World’

2026-04-21 · Source: AI & I - Every · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, extended

Summary

Kieran Klaassen, GM of Quora and creator of the Compound Engineering framework, discusses the "AI sandwich" metaphor for human-AI collaboration in a podcast with Dan Shipper. Klaassen explains that the core "work" phase of agentic coding is largely automated, allowing humans to focus on the initial brainstorming and ideation, and the final review and polishing stages. The Compound Engineering plugin, initially developed for Quora, incorporates planning, agent-driven work, review, and a "compound" step to integrate learnings. This approach, which applies to various knowledge work domains beyond engineering, emphasizes human involvement at the beginning and end of a workflow, where tasks require frame-setting, taste, and subjective refinement, while AI handles the execution of well-defined tasks.

Key takeaway

For AI Engineers and Product Managers designing workflows, you should adopt the "AI sandwich" model to optimize human-AI collaboration. Focus your team's creative energy on the initial ideation and final refinement stages, where human judgment and "taste" are irreplaceable. Delegate the repetitive, well-defined execution tasks to AI agents, freeing up human talent for higher-value, more enjoyable work that drives unique product differentiation.

Key insights

Humans excel at framing problems and refining outputs, while AI efficiently handles the execution of defined tasks.

Principles

• Humans set the problem frame.
• AI executes defined tasks.
• Human taste refines AI outputs.

Method

The Compound Engineering workflow involves human-led brainstorm/ideation, AI-driven planning and work, and human-led review/polish, with a compounding step for knowledge integration.

In practice

• Apply the "AI sandwich" model to your workflows.
• Focus human effort on initial problem framing.
• Dedicate time to polish AI-generated outputs.

Topics

• AI Sandwich Metaphor
• Compound Engineering
• Human-AI Collaboration
• Agentic Coding
• Knowledge Work Automation

Best for: Software Engineer, AI Engineer, Director of AI/ML

Related on AIssential

• Discipline is Taste — AI systems, optimized for plausibility and familiarity, degrade human judgment and homogenize content, creating a "discipline problem."
• Peter Yang on Small Teams, Coding Agents, and Why Human Ambition Has No Ceiling — AI agents are transforming software interaction and development, enabling radically smaller teams and automating knowledge work through conversational interfaces.
• The Hard Lessons from Running Dozens of AI Coding Agents in Parallel — Scaling AI coding agents amplifies single-agent failures, necessitating structured fixes and human strategic oversight.
• How We Bet Against the Bitter Lesson — Agent Skills integrate human expert judgment and deterministic tools to create a new, efficient AI-human knowledge economy.
• 😺 Anthropic: AI Is Building AI now — AI systems are increasingly handling execution tasks in their own development, fundamentally shifting human roles.
• Steve Yegge Wants You to Stop Looking at Your Code — AI agents will transform coding into orchestration, shifting human roles to managing complex problems and leveraging taste.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI & I - Every.