Skip to content

Knowledge base

Architecture and Design

Architecture matters more when agents write most of the code. The engineer's contribution moves to structure and judgment, and the artifacts that steer agents, such as specifications, tests, documentation and design systems, become parts of the architecture.

Written by

Updated:

Agentic Engineering 5

Our position. We hold that AI can generate code but cannot keep a growing system architecturally coherent (opens in a new tab), so someone has to own the system design before code is written. The codebase is the contract the agent works inside (opens in a new tab): tests, imitable naming and loud boundaries decide who gets value from AI, more than the model or the token spend.

The pages below carry that ownership into systems built from agents, the build-or-buy decision, the specifications agents read, legacy modernization, generated interfaces and video.

Start here

Resources

System Architecture

Documentation & Knowledge Management

UI Components & Design

In this section

  • Agent-authored video

    Agent-authored video treats a video as code: the agent writes a composition in HTML or React, headless Chrome renders it frame by frame, and FFmpeg encodes the result. The source diffs and reviews like code, but only watching the render tells you whether the video is good.

  • Architecting agentic systems

    An agentic system splits into agents and workflows instead of services, and connects them through MCP, agent-to-agent protocols and hand-offs that carry context. Production use needs durable execution, tracing, bounded loops, trajectory evaluation and a zero-trust posture toward every agent.

  • Architecture work in the agentic era

    When agents write most of the code, the engineer's job moves to architecture: placing boundaries, stating intent and constraints, and judging quality. Cheap implementation raises the value of design, because an agent without a sound design model only produces a mess faster.

  • Build versus buy

    Agents and English-language builders let anyone build software, but they do not make buying obsolete. Business software encodes rules most builders never see, so companies keep buying components, now designed to be driven by people and agents alike through protocols such as MCP.

  • Modernization and interface design

    Agents read undocumented legacy code without tiring, which makes reverse engineering a strong use case, as long as deterministic parsers and tests bound what the model touches. Generated UI needs the same discipline: without explicit design direction, models converge on a generic look.

  • Specifications, tests, and context

    The specification is the source of truth and its tests check the code against it. When the two drift apart, the spec's owner decides which side changes. Documentation now has a second reader, the agent, which needs a small, ranked and maintained instruction layer instead of an exhaustive one.