When people go on holiday, the knowledge stays in the building

At a glance

• Starting point: After nearly a decade of working together, knowledge about projects and decisions sat scattered across inboxes, spreadsheets and individual people's heads, and was hard to retrieve whenever a colleague was away.
• Solution: We built a central platform where projects, time tracking, documents and knowledge entries all sit on one data foundation, complemented by controlled AI analysis.
• Outcome: Answers that used to take hours are now available in seconds, running in production and daily use since early 2025.

The starting point

Over the years, a whole landscape of systems had grown up between us and a GDP-certified pharma logistics provider: web shops, business applications, interfaces and automations across a double-digit number of projects and several clients. Around 15 key account managers look after it.

With that depth came a quiet risk. Knowledge about projects, agreements and decisions sat exactly where it sits in most organisations: in inboxes, in spreadsheets and in individual people's heads. Three situations make that expensive, and every management team recognises them. The colleague responsible is on holiday for two weeks, a client asks for a project status, and nobody can answer. An experienced employee changes jobs, and their successor is briefed by colleagues who don't know everything themselves. A compliance question comes up, and the answer sits in an email thread nobody can find any more.

Our approach

The obvious answer would have been yet another tool. We built the opposite: one platform where everything converges. Projects, support hours, documents and project knowledge sit on the same data foundation, with the same clients, projects and permissions. No exporting between separate point solutions, a shared memory instead of scattered individual records. Only once that foundation was in place did automation and AI go on top, controlled and traceable.

The solution

• Central project and client management. Every project with status, quota usage, documents, knowledge entries and tasks on one detail page. The whole relationship at a glance.
• Time tracking with forecasting. Support hours by project, employee and category, with a live quota status and a projection that shows early whether the monthly budget will hold. Every invoice line traces back to specific entries.
• Knowledge base and versioned documents. Every project has its own wiki and a central, versioned file store. Who loaded which version, and when, is logged. No knowledge gets lost when a colleague takes over or a client asks a question months later.
• Audience-appropriate summaries at the push of a button. The platform analyses a project's actual status and drafts summaries pitched at the right level: for decision-makers, project leads or account managers. The AI proposes, a responsible person approves.
• An AI assistant that only reads. Questions in natural language, answers from our own project world instead of the internet. Deliberately separated from anything that writes, working only on approved data, with sources cited. It answers questions; it changes nothing.
• Tiered security model. Four confidentiality levels and separate visibility per item, confidential data individually encrypted, role-based access, hosted exclusively in Germany. Without clearance, you don't even see an entry in the list.
• Automatic, subscribable reports. What matters reaches the right person, instead of someone having to check dashboards. Strictly opt-in, every send logged.

A person deliberately sits at the start of every chain: only authorised people decide what goes into the system at all. That keeps the knowledge base free of noise.

Knowledge gained, time saved

This is where the real leverage sits. A question like "how much support time went into project X this month?" answers itself in ten seconds instead of half an hour of back and forth. If someone steps in for an absent colleague, the platform gives them an overview of a client relationship in minutes, work that used to take days of briefing. And the project knowledge builds up continuously, searchable and traceable, instead of being pieced together painfully after the fact. Scattered communication becomes a robust memory that holds up under scrutiny too.

The immediate benefit for the provider: their people can answer their own customers without having to ask us first.

The outcome

• The entire project and client landscape of a nearly ten-year engagement in one place, running in production and daily use since early 2025.
• Answers in seconds instead of hours, stand-in onboarding in minutes instead of days, knowledge preserved independently of any one person.
• AI where it safely holds up: read-only, on approved data, with a person in the decision loop. Fully traceable, hosted in Germany.

What this story shows

We don't just deliver individual systems, we think about the whole working relationship. Nearly a decade of growing projects produced a tool that protects knowledge and gives time back, because we built exactly what was missing between the existing systems and the people. It's the same discipline we use to build individual business applications, just applied to the whole relationship.

Running and developing a system for years is part of the partnership for us, not an add-on; how we organise that is on our pharma IT page.