Scale & Strategy
together with
This is Scale & Strategy, the daily newsletter that's crispier than the first chip out of the fab.
Here’s what we’ve got for you today:
- OpenAI's secret weapon underneath Codex
- AI just made the one-person unicorn a lot more realistic
OpenAI's secret weapon underneath Codex
The biggest story inside OpenAI right now isn't Codex.
It's the infrastructure underneath it.
If you follow AI, you've probably heard plenty about Codex, OpenAI's coding agent that's competing with Claude Code and has grown to roughly 10 million monthly users.
But the more important technology is the agent harness powering both Codex and ChatGPT Work.
"The harness is how the model interacts with the world and how we are able to express the model capabilities," said Joe Gershenson, who leads harness engineering at OpenAI.
Think of the harness as the conductor of an orchestra. It manages context, decides which tools to call, prompts the user when necessary, coordinates plugins, executes actions, and keeps the model moving toward completing a task. The model provides the intelligence. The harness turns that intelligence into useful work.
The downside is that all of that orchestration can burn through an enormous number of tokens.
Back in January and February, when OpenClaw and other autonomous agents first took off, developers were reporting token bills north of $20,000 per month because their agents repeatedly called models, tools, and APIs to complete complex workflows.
OpenAI has spent the last several months attacking that problem by optimizing the harness, the inference layer, and the API stack.
The payoff is significant:
- GPT-5.6 Sol (max reasoning) outperforms Claude Fable 5 on the Artificial Analysis Coding Agent Index while using 54% fewer output tokens.
- GPT-5.6 Terra delivers performance comparable to GPT-5.5 at roughly half the price.
- GPT-5.5 Luna is now OpenAI's fastest model while costing 80% less than Sol.
That's welcome news for anyone building agents, especially enterprises that have been watching inference costs spiral.
ChatGPT Work is essentially Codex for everyone else. Rather than targeting developers, it's designed to bring AI agents to the other 990 million ChatGPT users. By reducing token costs while making ChatGPT Work available from mobile and the cloud, OpenAI is betting that far more people will begin delegating long-running tasks to AI.
"I'd love to see people get more ambitious with their prompts," said Ahmed Ibrahim, a member of the technical staff at OpenAI. "Take a task that originally would take a day or two or a week, give it enough context and see how it works."
Those ambitions only become practical if agents remain affordable.
The timing couldn't be better.
Enterprise buyers have shifted from asking who has the smartest model to asking who can deliver intelligence without exploding the inference bill. Databricks CEO Ali Ghodsi recently summed up the mood: "It's the number one thing we're getting asked: 'How do we curb the cost but still invest in AI?'"
AI labs have largely responded in two ways.
The obvious approach is making tokens cheaper.
The harder approach is generating fewer tokens in the first place.
OpenAI appears to be betting heavily on the second.
Rather than relying solely on lower-priced models, it's making the software coordinating those models dramatically more efficient. That may end up being a far more durable competitive advantage than simply cutting API prices.
Codex is what everyone sees.
The harness underneath it may be what actually wins the enterprise.
Risk and regulations are ramping up, and customers expect proof of security just to do business. That’s why Vanta is a game changer.
Vanta automates your compliance and security processes in one AI-powered platform.
Whether you’re prepping for SOC 2, or running an enterprise GRC program, Vanta keeps you secure and your deals moving.
Companies like Ramp and Writer spend 82% less time on audits with Vanta. Faster compliance—and more time for growth.
Check out the on-demand demo to learn more. – trusted by top startups like Cursor, Linear, and Replit.
AI just made the one-person unicorn a lot more realistic
For decades, scaling a business meant hiring people.
AI is starting to break that assumption.
A growing number of founders are now building million-dollar businesses without hiring a single employee, using AI to handle everything from customer support and operations to marketing, software development, and administrative work.
One founder launched an AI software company just last December and is already on pace to generate roughly $10 million in annual revenue with 10,000 paying customers. His employee count? One.
He's far from alone. According to data from Stripe, the number of solo founders generating more than $1 million in annual revenue doubled between 2023 and 2025.
The implication is bigger than a handful of successful indie hackers.
For years, software allowed small teams to compete with large companies. AI is beginning to let individuals compete with teams.
Instead of hiring specialists for customer support, copywriting, operations, design, or even engineering, founders can increasingly stitch together AI agents that perform much of the same work at a fraction of the cost.
That doesn't mean employees disappear.
It does mean the threshold for building a meaningful company is falling dramatically.
Historically, reaching $1 million in revenue often required building an organization.
Today, it increasingly requires building a system.
Why it matters: AI isn't just making existing companies more productive. It's creating an entirely new category of businesses where one person can accomplish what previously required an entire team. If that trend continues, the biggest shift in entrepreneurship over the next decade may not be larger startups. It may be dramatically smaller ones.
Was this email forwarded to you?
That’s it for today and as always It would mean the world to us if you help us grow and share this newsletter with other operators.
Our mission is to help as many business operators as possible, and we would love for you to help us with that mission!
