Automating project status updates with the right tools and the right controls.
We built a governed reporting platform that combines Power Automate, SharePoint, Word templates and targeted AI. The outcome is faster reporting, stronger consistency and a reusable intelligence layer for future client services.
Built around practical governance
The point was never to use AI everywhere. The point was to use it where it adds value, inside a controlled process.
Selective AI adoption
AI is used where interpretation adds value, such as turning timesheet detail into a professional reporting snapshot.
Governance-first design
Workflow controls, permissions, approvals and retention are designed into the process from the start.
Human-in-the-middle controls
Consultants remain accountable for the final report. Automation drafts, humans validate.
SharePoint intelligence layer
SharePoint provided the right foundation for today: governed, secure and quick to implement. It also keeps the pathway open for more advanced reporting and analytics as the solution matures.
Future client interaction
The same knowledge layer can support governed AI-assisted responses to client questions.
Right tool for the job
Power Automate handles repeatable process. AI handles language and interpretation. People handle judgement.
Why not AI agents for everything?
Current AI hype suggests agents should be used for everything. In practice, this often leads to failed initiatives, uncontrolled AI consumption costs and increased operational risk when organisations don't implement the right governance and controls.
Reliable process automation
Processes such as report assembly, approvals and document distribution benefit from predictable, repeatable workflow automation that is easier to govern, support and scale.
AI where it adds value
AI was applied selectively to tasks requiring interpretation and language refinement, including converting operational activity into concise, professional client reporting.
Governance still matters
As agent autonomy increases, organisations often need additional controls such as monitoring, confidence scoring, escalation paths and validation workflows, which can significantly increase cost and complexity.
Use automation where precision matters, AI where interpretation adds value, and people where judgement matters.
Agentic AI can absolutely be the right solution in some scenarios, particularly where workflows require dynamic reasoning, orchestration and contextual decision-making. KJM also develops these agentic real-world implementations and will feature them in future case studies.
End-to-end workflow
Operational data retrieval
Power Automate retrieves project and timesheet information from the finance and operational systems.
SharePoint data layer
Project data is stored in SharePoint as a governed operational repository.
Carry-forward data
Status indicators, descriptions and continuing items are pulled from prior reports.
Report assembly
Word templates are populated with approved fields, including project health, budget hours, hours used and remaining balance.
AI activity snapshot
Timesheet descriptions are summarised into professional client language.
Human review
The consultant reviews and updates the draft where judgement is required.
PDF distribution
A SharePoint button creates the PDF and sends it to approved customer recipients.
Quality governance
A future AI quality check can escalate exceptions to senior review.
SharePoint: the reusable project intelligence layer
Aligned to Plan, Deliver, Grow
This is a practical example of KJM’s operating model applied to our own business.
The design decision that mattered
We deliberately avoided using AI for tasks better suited to reliable workflow automation.
We understand that AI bill shock is real, and in other cases aggressive throttling can impact operational processes and user experience. Our designs reflect today’s practical realities while remaining flexible enough to evolve as AI capability matures and consumption becomes more commoditised.
Future capability
Once the governed knowledge layer exists, the model can extend into more advanced AI-assisted service interactions.
AI-assisted client questions
For example, clients could submit project questions through email or a portal interface. A governed workflow could then validate the requester, retrieve approved project information, use AI to interpret the request and prepare a contextual response for review or controlled release.
Business outcomes
The value is not just automation. It is consistency, governance and a foundation for future intelligent services.
