Connect the document to the operation
Project documents should not become dead files after approval. Objectives, indicators, activities and requirements need relationships with programme implementation.
Development programmes often begin with detailed research, proposals, donor commitments and workplans.
But once implementation starts, those documents can become disconnected from day-to-day activities, field engagements and reporting.
This system was designed to keep the programme connected — from what was promised, to what was planned, to what happened, to what ultimately needs to be reported.
A development programme can be governed by multiple sources of information: research, proposal documents, donor agreements, objectives, workplans, indicators, reporting requirements, meeting notes, field engagements, participant records, activity evidence, follow-ups and periodic reports.
The problem is that these pieces of information can become operationally disconnected.
The proposal may define what the programme promised. The workplan may define what should happen. Field teams record what actually happened.
Reporting teams later need to reconstruct the relationship between all three.
The systems challenge was to connect these layers rather than digitise each one independently.
Business architecture — operational layers, not infrastructure.
Business architecture — operational layers, not infrastructure.
Programme documents contain information that should influence day-to-day implementation.
Instead of leaving that information buried inside files, the platform creates structured knowledge categories that can support programme operations.
The objective is not merely to summarise documents. It is to make programme knowledge available to the operational system.
These values represent information captured or extracted by the system from programme documents. They do not represent work the software performed on behalf of the programme.
A workplan should not remain a static document once implementation begins.
The platform models workplan items so planned activities can become trackable operational commitments.
The workplan captured is the Annual Workplan 2026 at version v1, with two items visible and an average completion of 0% at the moment of capture.
0% represents the state of the workplan when the screenshot was captured. It is not an indication of programme failure or system performance.
Activities provide the connection between the programme plan and work carried out by the organisation.
The captured system currently shows two planned activities, including a Policy Dialogue and a Training Workshop.
The activity layer can connect programme work with the wider operating model.
Not every engagement takes place behind a desk.
The system includes a field workflow designed around officers conducting meetings, dialogues and other engagements, supporting structured capture of implementation evidence.
Typed notes are supported where a meeting is not recorded, and the workflow is intended to remain usable while an officer is conducting an onsite engagement.
The field layer converts a planned activity into a recorded engagement with participants, notes and evidence attached to it.
Because the engagement is modelled rather than filed, follow-ups and activity evidence remain connected to the reporting layer.
Traditional reporting workflows often require staff to reconstruct activity information after implementation has already happened.
This platform is designed so reporting draws from structured programme data captured throughout the operating cycle.
The intention is not to have AI invent reports from nothing. The system provides structured evidence and programme context from which reporting can be produced, reviewed and approved.
The Activity Report is the first reporting layer. Each subsequent stage represents a different level of programme accountability.
Producing a report internally is not necessarily the end of the reporting workflow.
Where donor submission is required, the system needs visibility into the submission requirement, the deadline, report readiness, submission status, donor response, required actions and follow-up.
These are point-in-time operational states captured from the system, not a measure of donor compliance.
The platform includes an “Ask Project Intelligence” capability. Its purpose is to allow users to interrogate processed programme documents without separating the AI experience from the project's source material.
The interface explicitly states that answers should be generated only from processed project documents, and should cite the relevant page, sheet or section.
This creates a fundamentally different experience from using a generic AI chatbot. The assistant operates inside the programme's knowledge boundary.
In organisational environments, a plausible answer is not enough. Users need to understand where important information came from.
The system is therefore designed around source-grounded project intelligence rather than unrestricted generation.
What indicator did the donor agreement specify?
What activity was included in the approved workplan?
What deadline applies to a reporting requirement?
What did the research identify?
AI should reduce the time required to find institutional knowledge without removing the ability to verify it.
Once implementation data is structured, management can understand programme activity without manually rebuilding the picture from separate documents and spreadsheets.
These are values from the programme at the moment captured. They are not final programme outcomes.
The executive view should surface programme state and exceptions without requiring management to inspect every underlying record.
Every value reflects the programme at the moment captured, not a final programme outcome.
Project documents should not become dead files after approval. Objectives, indicators, activities and requirements need relationships with programme implementation.
Reporting becomes easier when programme evidence is structured as activities occur rather than reconstructed months later.
Field officers need a simple path through participants, notes, evidence and follow-ups. The system must support typed notes where audio recording is not used.
Activity, monthly, mid-year and quarterly reporting represent different stages of programme accountability.
AI becomes more useful when it operates within the project's knowledge boundary and provides a path back to the source.
A donor requirement should not be remembered only when the submission deadline approaches. It should exist as part of the programme's operating model.
“Treat commitments, indicators, activities, evidence and reports as one connected information model — then layer AI on top with source traceability.”
My role centred on understanding how programme knowledge, implementation and reporting fit together and translating that operating model into a connected digital system.
“The most useful AI feature wasn't a chatbot. It was connecting AI to the knowledge the organisation already had — and preserving the path back to the source.”
Project Intelligence demonstrates that systems thinking is not limited to transactions, inventory or manufacturing. Knowledge itself can become operational infrastructure.
Beyond these three detailed case studies, I have designed and implemented systems across government, commerce, field operations, education, service businesses and internal operations.