# Your Business Is an Algorithm You don't need to become an AI expert. You need to keep becoming an expert in your business, because that knowledge is what software is built from. Everywhere I look, someone is telling business owners they need to become AI experts. AI is changing everything, the message goes, and anyone who doesn't catch up will fall behind. I think that gets it backwards. AI is a remarkable development. It doesn't follow that every business owner needs to become a technologist. You don't need to become an AI expert. You need to keep becoming an expert in your business. The thing that makes your business valuable isn't the technology you use. It's what you know how to do with it. The more I work with software and AI, the more that knowledge looks like the scarce ingredient. That's because your business is already an algorithm. ## Your business is an algorithm I've thought about businesses this way for a long time. Every business can be described with a simple model: **Inputs → Value creation → Outputs** - **Inputs** might be customers, enquiries, materials, information, labour, expertise and capital. - **Value creation** is the decisions, activities, transformations and interactions that turn those inputs into something worth more. - **Outputs** might be a completed project, a resolved problem, a delivered product, a customer experience or a measurable improvement. Money is an important output. It's also the return that lets the business keep producing the outcomes it exists to deliver. Real businesses are more complex than three boxes. They have interconnected processes, feedback loops, dependencies and conditions that change. So this is a way of looking at a business, not a claim that a business reduces to a formula. It's useful because it tells you where to point your attention. The particular way you arrange those activities is a large part of what makes your business different. Michael Porter made a version of this point in [What Is Strategy?](https://hbr.org/1996/11/what-is-strategy): "The essence of strategy is choosing to perform activities differently than rivals do." ## What is your algorithm trying to achieve? Every algorithm needs a defined output. Businesses are no different. A business that optimises for revenue alone may make very different decisions from one that puts a particular customer outcome first while staying financially sustainable. That isn't a moral judgement. It's a question of internal clarity. When the people inside a business understand what they're working towards, they share a basis for making decisions. The objective can be complex, and it can change. It doesn't need to be a tidy mission statement. It needs to be understood. Take music as an illustration. A business focused on selling CDs may find it harder to see the opportunity in making music more accessible. A business focused on helping people discover and enjoy music has more freedom to change how it delivers that. That's a simplification of a long story. I use it only to show how an objective shapes the processes a business builds and the opportunities it notices. ## You've already been writing the algorithm For decades, well-run businesses have worked to standardise their operations. They write procedures, document workflows, assign responsibilities, set quality standards, train people and measure results. Each of those is an attempt to describe the logic of the business. A standard operating procedure (SOP) might define: - what triggers a process - what information is needed - what decisions must be made - what activities need to happen - what a successful outcome looks like That is remarkably close to what you'd need to specify a software workflow. An SOP isn't a complete software specification. As I wrote in [Business as code](/blog/business-as-code/), a written procedure still depends on who interprets it and how often it's updated. Software needs the exceptions and judgement calls spelled out too. But an SOP is an excellent starting point. You've already been doing much of the groundwork for software automation. You just called it running a business. ## What's changed is the economics of software Turning a business process into custom software has historically been expensive. It took specialist knowledge, time and ongoing investment. So plenty of smaller operational problems were never worth automating. The software would have cost more than the problem. AI-assisted development is starting to change that equation. It can speed up parts of programming, prototyping, testing and implementation. The gains vary by task. In a [2023 controlled experiment](https://arxiv.org/abs/2302.06590), developers using GitHub Copilot built a small web server 55.8% faster than those without it. In a [2025 study by METR](https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/), experienced developers working in their own large codebases took 19% longer with AI tools. METR has since reported that the tools have improved, though its newer data can't yet say by how much. My reading is that the better defined the task, the more the tools help. That matters for what comes next. It opens room for smaller, targeted pieces of software built around specific processes. More businesses can consider building software around how they operate, rather than always adapting their operations to fit generic software. Not every business needs custom software. Off-the-shelf tools are often the right choice. Requirements, security, integration, testing and maintenance still matter too. That's where software engineering expertise earns its place. The revolution isn't that businesses suddenly need algorithms. They've always had them. The revolution is that turning those algorithms into software is becoming more accessible. ## You're closer to AI than you think So you don't need to abandon the operational improvement work you're already doing. You don't need to master every new AI product. It's also worth remembering that deep knowledge of AI doesn't automatically mean someone understands your business. Keep doing what good businesses have always done. Define processes. Improve procedures. Identify problems. Measure outcomes. Understand customers. You know why a customer calls. You know what a successful job looks like, where a process tends to stall and what makes your business different. That knowledge decides what the software should do. AI can help with implementation and interpretation. It doesn't supply the operational judgement and experience behind the requirements. You can hire someone to write your software. You can't hire someone to have your experience. ## Start with one small algorithm Pick one process where the inputs, decisions and outputs are reasonably well understood. Handling an incoming customer enquiry is a good example. - **Input:** an enquiry with the customer's details and a request. - **Process:** identify the customer, categorise the request, decide who should respond and record the action needed. - **Output:** an assigned task, an acknowledgement and a response you can trace. That's a small business algorithm. It might start out handled by hand, then get written up as an SOP, then run through conventional automation or software. AI could help interpret an ambiguous email or pull details out of an unstructured request. It isn't needed for every step. A sensible progression: **Understand → Document → Standardise → Improve → Automate → Measure** You can improve the process in stages. Nothing has to change all at once. ## Keep becoming an expert in your business Technology will keep advancing. New tools will appear, and software development will keep changing. You don't need to chase every development. Don't let the noise around AI distract you from what you already know. Keep improving your processes. Keep documenting how your business works. Keep understanding your customers and what makes your business valuable. The better you can describe the work your business does, the easier it becomes to turn that work into software. As the cost of building software falls, that knowledge becomes more useful. These are ideas I'm still testing. Next, I want to check them against the research on strategy, software productivity and what businesses are getting from AI. You don't need to become an AI expert. You need to keep becoming an expert in your business. ## Sources - [What Is Strategy?](https://hbr.org/1996/11/what-is-strategy) — Michael E. Porter, Harvard Business Review, 1996 - [The Impact of AI on Developer Productivity: Evidence from GitHub Copilot](https://arxiv.org/abs/2302.06590) — Peng, Kalliamvakou, Cihon and Demirer, 2023 - [Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity](https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/) — METR, 2025