Picking the right AI project is the easy part. Getting it into production, and keeping it there, is where most of the money goes missing. A good backlog is worth very little if the first item stalls between the pilot and the people who are supposed to use it.
Earlier this month we explained SCALE, the five conditions an AI effort needs to succeed. This is the other half: PROPEL, how we carry one opportunity from the backlog to something the client owns.
What is the PROPEL framework?
PROPEL is how Pisteyo delivers. It takes one prioritized opportunity through execution, proof and scale, and it ends with the client owning what was built. Each step answers one question, and each produces something you can check.
| Step | The question it answers | What it produces |
|---|---|---|
| P: Prioritize | Which opportunity goes first? | A ranked choice, weighed on impact, user pain, data readiness and time to value |
| R: Resource | Who owns it, and can we get to what we need? | A named sponsor, owner and team, with access and decision rights confirmed |
| O: Operate | Is it working for the people who will use it? | Short build cycles, real demos and feedback from end users |
| P: Prove | Did the work actually change? | Measured adoption, time saved and errors removed, reported apart from estimates |
| E: Expand | Where does it go next? | The proven solution in the next team and workflow, with governance to match |
| L: Lead | Can the client run it without us? | Trained owners, documentation, governance and internal champions |
Why do AI projects stall after they are chosen?
Because the hard part starts after the decision. A project gets picked, a pilot gets built, and then it waits: for a permission nobody requested, for feedback from users who saw it for the first time at the end, for a result nobody measured, for an owner who was never named. None of those is a model problem. Each is a delivery problem, and each one has a step in PROPEL.
What are the six steps of PROPEL?
P: Prioritize
Not everything on the backlog deserves to go first. Choose the opportunities that best balance business impact, user pain, data readiness, technical complexity, security and compliance, change impact, time to value, and how measurable the result will be.
In practice: weigh impact, user pain, data readiness and time to value before anyone builds. The first project should be one that can prove itself quickly and visibly, because early wins are what earn the second project.
R: Resource
Build the right team before the build. Confirm access, ownership, capacity and decision-making authority first. Depending on the work that means an executive sponsor, a business owner, a delivery lead, subject-matter experts, a change lead, engineering, and data and security stakeholders.
In practice: confirm the sponsor, the owner, the access and the decision rights before the build starts. The delays we see most often are a missing permission, not a missing feature.
O: Operate
Speed with transparency. Short cycles, clear backlogs and acceptance criteria, demonstrations and feedback loops, risk and dependency tracking, documentation and quality controls, with training and change work running alongside the build rather than after it.
In practice: short cycles, real demos, and feedback from the people who will use it. Test with the end user, not at the end.
P: Prove
Validate that it works and that it creates value. That means functional acceptance, actual user adoption, quality measures, time saved, reduced manual effort, better throughput, fewer errors, and evidence that the workflow itself changed. This is the step where estimated value becomes realized value.
In practice: measure what actually changed: adoption, time saved, errors removed. Estimated value and realized value are never reported as the same number.
E: Expand
Scale what has been proven, not what has been promised. More departments and user groups, more workflows, deeper integrations, reusable components, broader training, and the governance to support it all at a larger size.
In practice: scale only what has proven itself, into the next team and the next workflow.
L: Lead
Hand it over, and build the capability to keep it going. Knowledge transfer, operating manuals, client-owner training, support and escalation processes, governance handoff, backlog transfer and internal champions, so the client can operate, govern, improve and extend what was built.
In practice: training, documentation, governance and internal champions. The engagement isn't done until the client can run it without us.
The engagement isn't done until the client can run it without us.
How do SCALE and PROPEL fit together?
SCALE creates the conditions; PROPEL turns them into outcomes. SCALE gives an organization a clear strategy, readiness for change, the right tools, engaged leadership and an educated workforce. PROPEL converts those conditions into delivered results, one opportunity at a time.
The join between the two is disciplined discovery and scoping. That is where strategic intent becomes a small number of clearly defined initiatives, each with an owner, a measure, its dependencies and an agreed definition of what proof will look like. Skip the join and you tend to end up with a strategy nobody executes, or a build nobody asked for.
Which of the six does your team usually skip?
Most stalled AI work skips one step, and it is usually the same one every time. Six questions show which.
- Prioritize. Was the first project chosen on impact and readiness, or because it was the most exciting demo?
- Resource. Were the sponsor, owner, access and decision rights confirmed before the build started?
- Operate. Did the people who will use it see it in the first two weeks, or at the end?
- Prove. Can you show what changed, separately from what you expected to change?
- Expand. Is the next rollout based on proof, or on enthusiasm?
- Lead. If the project team left tomorrow, who would run it?
The step you hesitate on is the one to fix first.
Where should you start?
With one opportunity and an honest look at which step usually breaks. The AI Executive Briefing is a free strategy session for your leadership team, built around your industry: where your sector actually is with AI, the paths open to you, what your board needs to know about governance, and a starter framework you can use straight away. It is not a pitch, and you leave with a clearer view of where your AI work stalls, whether or not you work with us.
Book an AI Executive Briefing, and bring the step your team skips. If you have not read the first half of this series, start with What Is the SCALE Framework?