AI Strategy Guide: Specific Strategies for AI Growth

AIcan double the productivity of the people at your company. That is a tremendous prize and a large amount of change. Like anything big, it requires thoughtful consideration of the effects and sequencing. This guide will take you through how to achieve these results and how to roll out this transformation. Crucially, it is a plan for how to do so while building up the economy and society.

In this guide we focus on specific changes to make and specific patterns to build. We will fill in the whole picture but we’ll start with getting these techniques into your hands.

FAQ

Frequently asked questions

What is worth building vs. what can be done in Claude CoWork or similar computer-use programs?

The core principle here is that we want generative power and deterministic structure. Our staff should be able to Create something new... and know that systems and rules will be followed: safety, communication, approvals, budgets, security, data controls, laws. More (reads the full answer in the guide, opens in a new tab)

Build vs. buy in 2026

The economics of building software have changed! And it's still a lot of work to build good software. It is still worth buying good SaaS solutions. More (reads the full answer in the guide, opens in a new tab)

Build internally or externally?

Some of each. Internal teams have context and relationships. External teams have focused on build efficiency and have cross-company context. More (reads the full answer in the guide, opens in a new tab)

How can we be secure against AI-powered hacking?

It's hard. Companies need to seriously level up their cybersecurity. The main answer is that we all will need to spend more over the next few years both on internal capacity and on vendors. More (reads the full answer in the guide, opens in a new tab)

Introduction

Most companies use AI to change their business in three ways.

Research AI, building new models and systems to create new knowledge (e.g. AlphaFold).
Domain AI, AI services driven by key data that your processes accumulate (e.g. pricing models).
Workflow AI, quickly creating software to manage day to day tasks.

The first two get the most attention but Workflow level AI is widely applicable and hugely valuable. Rather than requiring hard to find ML engineers, much workflow improvement can be done by staff who currently do the work and by engineering staff you already have or already work with. The frequent complaint, “I wish this task was easier”, can instead become a tool for frequent AI delegation.

“How?”, you may ask. Software has undergone an AI revolution. This change is clearly visible and measurable. As a result the economics of software have changed. The difference between a software algorithm and a business process is wafer thin. This is what makes Workflow AI ripe for improvement and software leverage should be the launchpad for attacking Domain AI and Research AI projects as well.

A small outline figure stands before a signpost where a single path splits into several diverging paths drawn in white line work on a deep navy field: one doubles back on itself, one coils into a loop, one climbs straight up. One path onward is lit in amber.

This journey is going to get weird; and it will take us to great places, where we have so much capability at our fingertips. Imagine everyone having their own personal construction company. It’s one thing to know that abstractly, it’s another to see the office parking lot when everyone drove an excavator to work. Working for technology companies, our team has been through this transition first hand. The job of software engineers writing code went away last year (even if many don’t know it yet). The new job of software engineers crafting outcomes is fulfilling and wonderful.

Many AI prescriptions suffer from “software brain”, the perspective that everything is work to be automated. Or “business brain”, focusing on short term profits over healthy systems. This guide does focus on automation and creating business value but it only works if we first align on where we are headed both personally and professionally.

Several separate paths drawn in white line work on a deep navy field, each with a small outline figure walking along it, run in from different directions and merge into a single path. The combined path is lit in amber from the point where they meet.

Getting to success all together as companies and societies means talking through, “how to share the value we create”, and “how to avoid externalities”. We can’t lay everyone off then wonder why no one has money to buy our products. We can’t expect our coworkers to drive change unless that change is safe and the incentives work for them instead of against them. We have to purposefully choose paths that lead to growth, broad-based success, and good long-term outcomes.

Dividing up work between AI and people compounds. As tasks get fully or partially automated, new opportunities are revealed. The core work of finding constraints and imagining solutions becomes clearer across roles. This is not just an initial guide but a roadmap covering intermediate and advanced stages: how to drive demand and how to succeed along multiple dimensions. Market value, rewarding jobs, a booming economy, human flourishing; all of these are our goals.

What’s required to move forward? Commitment. Start the alignment conversations, allocate a relatively small budget for the first project, commit to transforming your company over the next two years.