When an AI project stalls, it's usually not the tech. It's who's doing what.
Nina Brenes··6 min read
- Every AI project has five roles: AI Visionary, AI Operator, AI Implementer, Subject Matter Expert, and End Users.
- Mixing the roles up, not the technology, is the most common reason AI projects fail.
- The AI Operator translates processes into playbooks and drives adoption. It is not technical, and it is the role most consultants should play.
- The five-role model is from the AMP program (runonamp.com). At The Harmony Labs, Governance is added as a sixth discipline.
Five roles. One project. Each role has a distinct job. When someone plays the wrong role, the whole project stalls. And that, more than the technology, is the number one reason AI projects fail.
And projects failing is not rare. RAND found that more than 80% of AI projects fail, about twice the rate of ordinary IT projects (RAND Corporation, 2024), and S&P Global reported the share of companies abandoning most of their AI initiatives jumped from 17% in 2024 to 42% in 2025 (S&P Global Market Intelligence, 2025). This five-role model is from the AMP program (formerly The AI Exchange), where I certified as an AI Operator. It is not theory: mixing these roles up is the most common mistake I see behind those numbers.
The five roles
1. AI Visionary
Sets the why, the priorities, and the budget. Their job is to define what success looks like, then stay out of the weeds. Not involved in the build.
2. AI Operator
Translates business processes into playbooks, tests the system, and ensures adoption. This is the role most coaches and consultants should be playing. Not technical. Essential.
3. AI Implementer
Builds the technical system: writes code, configures tools, connects integrations. Works from the playbook the Operator provides, and should not be designing the process, only building what has already been designed.
4. Subject Matter Expert
Defines what good looks like. They provide the expertise that makes the output standard-worthy. If you are automating client reports, the SME is the person who knows what a great client report contains.
5. End Users
The people who actually use the system every day. If they do not adopt it, nothing else matters. Adoption is the only metric that counts at the end.
How it breaks: playing the wrong role
Getting the roles wrong is almost always what stalls an AI project. The same four failure modes show up again and again:
- The Visionary tries to do Operator work, so it stalls and nothing gets built.
- The Implementer owns the process mapping, so they build the wrong thing.
- The SME tweaks prompts instead of defining standards, so the output stays inconsistent.
- Too many end users shape the system, so it gets bloated and unusable.
Where Governance comes in
The five-role model is from AMP (runonamp.com), the program where I certified. At The Harmony Labs I add a sixth discipline, Governance: data protection, responsible-use guardrails, and a human in the loop where it matters, owned from day one. Know which role you are playing on any given project, and stay in it.
FAQ
What are the roles in an AI project?
Five: the AI Visionary (sets the why and budget), the AI Operator (writes the playbooks and drives adoption), the AI Implementer (builds the system), the Subject Matter Expert (defines what good looks like), and the End Users (whose adoption is the real metric). It is a model from the AMP program.
What does an AI Operator do?
The AI Operator translates business processes into playbooks, tests the system, and drives adoption. It is deliberately not a technical role, and it is the role most consultants and coaches should be playing on an AI project.
Why do AI projects fail because of roles?
Because people play the wrong one. A Visionary doing Operator work stalls the build, an Implementer mapping the process builds the wrong thing, an SME tweaking prompts leaves output inconsistent, and too many end users make it bloated. The technology usually is not the problem.
Who should lead an AI project in a company?
The AI Operator runs the project day to day, while the AI Visionary owns the why and the budget from above. The Operator is the pivotal role, and it is non-technical, which is why an outside operator often fills it best.
- AMP (formerly The AI Exchange), Certified AI Operator program — AMP
- The Root Causes of Failure for Artificial Intelligence Projects (RRA2680-1): over 80% of AI projects fail — RAND Corporation
- Generative AI shows rapid growth but yields mixed results (Voice of the Enterprise: AI & ML 2025): abandonment rose from 17% to 42% — S&P Global Market Intelligence
Nina Brenes
AI partner for purpose-led, human-first founders
Certified AI Operator (The AI Exchange) and Anthropic Claude Partner. Nina spends her own time in AI conferences and daily practice so the founders she works with don't have to.
Certified AI Operator · Anthropic Claude Partner
Connect on LinkedIn