• Home
  • Blog
  • Projects
  • Bookshelf
  • About
← Back to all status updates

The Harness Isn't Your Problem, The Router Is

Published ....
Last modified ....
From Boston, MA

Share this status update on BlueskySee discussion on Bluesky


A lot of engineering work over the past year or so has been around building better harnesses. While that effort has brought a lot of great advancements (just look at the Codex ChatGPT mac app and all the features it supports for an example!)

That being said, I think the environment is shifting rapidly from one where the harness is all that matters to one where the router will be far more important.

The frontier models are really, really good. Yet each model can accel at different aspects of the problem being solved. Additionally, it can become cost prohibitive to send off that "fix this typo" prompt over to Fable 5 max each time. That's where a router comes in!

A router (at least in the context of AI agents) is a networking component that allows for different agents/models to be used for different requests. Imagine you send a prompt into the router, and then the router splits the work up by sending off a part of the prompt to a planning agent, and then takes the result of the plan and hands it off to a fleet of execution agents.

While you can use one model for everything (which I've been doing with GPT 5.6 Luna because of its cost efficiency), the best possible results for more problems is to leverage different models for different parts of the overall work.

A common breakdown I see shared widely on twitter at the moment is:

  • Use Fable 5 for planning
  • Use GPT 5.6 Sol for execution
  • Use GPT 5.6 Luna for sub-agents

There's a variety of hosted router companies out there, OpenRouter perhaps being the largest and most well known at the moment. However I think the next big leap will be an on-device router, e.g. something that sits on top of or next to your desired harness and handles routing to the different agents for you, without you needing to completely move over to only using a single global router.

Imagine for a second that you had a plugin (for lack of a better term) that integrated with your harness, analyzed the prompt you sent (maybe by a small and cheap model) and determined how to split up your prompt and send it off to different models and then stiched the responses back together.


Tags:

AISoftware Development

Loading...

Related Posts

AI

OpenBlank

....

ASD-STE100 Simplified Technical English

....

Performative Engineering

....

The act of making noise and delivering output instead of outcomes

The Agentic Engineering Scale

....

A brief outline of how I think of the scale of adoption of AI in software development

Onboarding Your New AI Teammate

....

Are we reinventing the wheel when it comes to onboarding new AI agents to a codebase, when we already have primitives available for onboarding humans?

Cursor and Obsidian

....

A quick look at combining Cursor with Obsidian to ask questions of your notes!

The AI Development Conundrum

....

Is it a good or a bad thing to offload writing code to AI agents and Large Language Models?

Offload Complexity, Don't Offload Learning

....

A rough mental model for how you should be leveraging AI as a tool for your own growth

My Current AI Stack

....

I've been using a variety of AI tools as of late, I figured I'd document the ones I'm primarily using!

Software Development

Performative Engineering

....

The act of making noise and delivering output instead of outcomes

The Agentic Engineering Scale

....

A brief outline of how I think of the scale of adoption of AI in software development

Onboarding Your New AI Teammate

....

Are we reinventing the wheel when it comes to onboarding new AI agents to a codebase, when we already have primitives available for onboarding humans?

Polish is Important

....

You're Building Software Wrong

....

Slicing software: why vertical is better than horizontal.