Executive Summary
AI is now a leadership decision.
AI is changing how companies compete. Competitors are using it to quote and price faster, serve customers at lower cost, reach markets they could not serve profitably before and offer services that did not exist a few years ago. A company that falls behind will struggle to catch up. The companies ahead keep learning from their own data and their people get better with the tools every month, while their customers come to expect the new level of service as standard.
Most companies have responded by letting AI spread team by team. Many of those agents and pilots produce useful results, and often very little of the gain shows up in the financial statements. Work built around separate priorities, processes and systems is hard to connect and harder to scale, while spending keeps growing and boards are starting to ask what the company has bought.
In our experience, most AI programs stall at the leadership level. Leaders have not agreed on what they want AI to achieve, and no single executive owns the results that depend on several departments. Deciding how the company will compete is the core job of the board and the CEO, and AI now sits at the center of that decision.
The three AI ambitions
The board and executive team first decide which ambitions AI will support. Octant groups them into three types, and each one asks something different of the company and its people.
| Ambition | Example | What it requires |
|---|---|---|
| Reinvent | A manufacturer that uses AI to predict equipment failures and sells guaranteed uptime as a subscription. | New skills, research into whether customers want the offering and, in some cases, permission to use customer data. Often funded in stages. |
| Expand | A bank that uses AI to assess small-business credit quickly enough to lend in new markets. | Evidence that customers will buy, a sales approach that fits them and employees trained to serve them. |
| Improve | AI that reduces errors and rework, improves service quality or speeds up quoting, claims processing and scheduling. | Protected time for employees to test the tools, review the output and change how the work is done across the teams involved. |
What leaders need to do
Set the ambitions and review existing work against them
Tie every significant AI project to an ambition and a business result. Existing agents and pilots get the same review as new proposals, and work that fits none of the ambitions is combined, kept small or stopped.
Name an owner who can deliver each result
When a result depends on several departments, one executive owns it across all of them. The CEO should confirm that executive has led change across departments before and has the team and time the work needs.
Plan for the dependencies projects share
Data, platforms, security and legal reviews, vendor contracts and people’s time each need an owner, a budget and a date, and every pilot that works needs a date for the decision to expand it.
Report business results against their full cost
Measures should fit each ambition and start from a baseline, with reviews at least quarterly. Counts of agents deployed or pilots completed only help explain the results.
Give good local work a route to scale and tell employees what is decided
Employees register experiments and link them to an ambition, and the responsible executive decides whether to expand, combine or stop them. Employees hear what has been decided, what is still open and who will decide it.
Questions for the next board meeting
The paper includes six questions directors can use to test whether the company has an executive mandate for AI, starting with which ambitions the company’s AI spending is meant to support and who owns each result that depends on more than one department. It ends with a one-page checklist for writing the mandate.
Topics
The full paper covers the three AI ambitions, the board’s and CEO’s roles, how to build the execution plan, how to give employee ideas a route to scale, six questions for the board and a one-page checklist for the executive mandate. Free, no gate.