Project success depends less on the initial idea than on the ability to structure each stage and stay on course when the unexpected happens. With the growing adoption of artificial intelligence in project management and new European regulatory requirements, the factors to consider from the outset of a project have multiplied. What separates practices that work from those that lead to projects stalling?
AI and project management: rapid adoption, benefits to be measured
The integration of artificial intelligence into project management has accelerated significantly. According to a 2025 survey by the Association for Project Management, cited by MeisterTask, 70% of UK project professionals said their organisation used AI, up from 36% two years earlier.
This growth does not mean the benefits are automatic. The same source notes that useful capabilities are available, but remain more limited than some sales pitches suggest. AI can speed up workload estimates, detect risks in a schedule or automate meeting notes. It cannot replace human decision-making when political or budgetary trade-offs are required.
| Indicator | Two years before the survey | 2025 survey (APM/MeisterTask) |
|---|---|---|
| Organisations using AI in their projects | 36% | 70% |
| Perception of AI capabilities | Little documented | Useful but limited |
The gap between adoption and actual maturity is what needs attention. A project manager who deploys an AI tool without defining roles, input data and validation criteria risks making the process more complex rather than simpler. Several resources can help structure this thinking, and reussi.fr offers practical guidance for approaching each project phase methodically.
Regulatory compliance: the AI Act timeline for projects
AI in a project is no longer just a matter of productivity. The European AI Act sets out a phased timeline that anyone leading a project needs to understand if they use algorithmic tools.
- Since 2 February 2025, provisions on AI literacy have applied, requiring organisations to train teams and document how AI is used.
- Since 2 August 2025, obligations for general-purpose AI models have been in force, including transparency and risk-management requirements.
- Subsequent deadlines vary by system type: high-risk systems are subject to stricter obligations, with deadlines extending beyond 2025.
For a project that uses AI, this changes the picture from the design phase onwards. Identifying the tools in use, the organisation’s roles and its legal responsibilities becomes a task to include in the project plan, alongside the budget and schedule.
Planning methods: what distinguishes a reliable plan from a fragile one
A fragile project plan has one telltale symptom: it cannot withstand the first unexpected setback. The difference between a reliable plan and a fragile one rarely comes down to the tool used. It comes down to task granularity and the quality of risk identification.
Task granularity and realistic estimates
Breaking a deliverable down into tasks lasting two to five days makes it possible to spot a delay in the first week. A three-week work package, by contrast, can conceal slippage until the deadline. Task granularity determines how early you can respond.
Estimates are more reliable when they come from the person who will do the work, rather than their line manager. This simple principle is still rarely applied in organisations where schedules are drawn up in advance and then imposed on teams.
Risk register and stakeholder communication
A risk register is only valuable if it is reviewed and updated regularly. Best practice is to revisit it at every milestone, involving the relevant stakeholders. An identified risk that is never reassessed is just as dangerous as one that is ignored.
Communication plays a direct role in risk management. A project manager who informs stakeholders of a deviation as soon as it is detected preserves their trust and ability to respond. A project manager who waits until the next steering committee meeting loses both.
Team skills and change management
Technical skills alone are not enough to guarantee a project’s success. A team’s ability to adapt to change has just as much, if not more, impact on the final outcome.
Managing change means anticipating resistance before it becomes a roadblock. This involves three practical steps: explain the project’s “why” to those affected, give them the means to adapt (training, tools and support), and measure their buy-in over time rather than waiting until the end.
A high-performing project team brings together complementary profiles, not only in technical skills but also in their approach to uncertainty. Pairing a methodical person with someone more exploratory helps ensure both adherence to the plan and the ability to spot unexpected opportunities.
Project tracking tools: choose based on actual needs
The project management tools market is vast. Spreadsheets, dedicated software, collaborative platforms with built-in AI: the right tool is the one the team actually uses, not the one with the most features.
A small project with three people does not need expensive licensed software. A shared tracking sheet with clear columns (task, owner, deadline, status) covers most needs. By contrast, a project involving several teams and complex dependencies requires a tool that can manage critical paths and flag cascading delays.
The deciding factor remains team adoption. A sophisticated tool that nobody updates provides less visibility than a simple file filled in every day.
The difference between a completed project and an abandoned one comes down to a few practices applied consistently: finely broken-down tasks, risks reassessed at every stage, direct communication with stakeholders and tools suited to the team’s actual context. The arrival of AI and the requirements of the AI Act add another layer of complexity, but also an opportunity to professionalise project management from the outset.
