How an AI project with Hoch-AI works

The process begins with a no-obligation introductory meeting. The goal is to understand the company, its current situation, and its most pressing challenges.

Typical topics include the starting situation – where do recurring manual tasks arise today? –, the desired outcome, the affected user groups, the technical environment including existing systems and Microsoft 365 structures, and the question of economic viability.

The goal of this meeting is not to sell a ready-made solution right away. Far more important is an honest assessment of whether an AI project makes sense here and what a useful first MVP might look like.

When the direction is right, the specific use case is sharpened together. This typically takes two to three meetings.

These sessions clarify the most important foundations: concrete goals, affected processes and users, relevant data sources such as documents, emails, or SharePoint content, measurable success criteria – and deliberate scope limits, meaning what should not yet be part of the first step.

The last point is particularly important. Good AI projects rarely fail because there are too few ideas. More often, the first step is scoped too broadly. For this reason, the use case is cut so that a realistic MVP emerges – one that delivers value quickly while remaining expandable.

Based on the worked-out requirements, a proposal is created. It describes the goals, the MVP scope – which features, user groups, and data sources are included in the first version –, the approach, the expected result, and the planned investment.

The proposal creates a clear framework without executing the implementation rigidly. AI projects thrive on learning during execution. Some assumptions are confirmed; others need adjustment. For this reason, we work in an agile manner and regularly assess which priorities currently deliver the greatest value.

During the implementation phase, the first usable version is built, tested, and improved step by step. Depending on the project, this could be a Teams knowledge bot, an Outlook assistant, or a process automation.

What matters is that the MVP is practically usable as early as possible. This makes it visible not just in theory, but in practice – whether the solution helps in daily work, what questions arise, and where it needs refinement.

Throughout implementation, there is a weekly sync. This meeting reviews current progress, collects feedback on what already works, clarifies open questions and access needs, sets the next priorities, and surfaces technical or organizational risks. This prevents any black-box effect – the client sees early how the solution is developing, and feedback can flow in directly.

At the end of the MVP or pilot project, the results are assessed together. This is not just about whether the solution works technically. What matters is whether it is practically useful in everyday work and whether further development is worthwhile.

Typical questions include: Were the defined goals achieved? Did the MVP deliver real value? Where is the greatest benefit? What should be improved? Which user groups could follow next? What support makes sense in ongoing operations?

Based on these answers, a follow-up proposal can emerge – for example, further development at a clearly defined fixed price for additional features or new user groups, or a service agreement for technical support, quality assurance, and regular optimization.

The approach combines structure with flexibility. There is a clear goal, a defined scope, and transparent costs. At the same time, there is enough room to respond to new insights as the project progresses.

This is especially important for AI projects. Real-world testing with actual data and actual users often reveals which features are truly valuable – and which are not.

That is why the starting point is not a large, abstract transformation program, but a pragmatic MVP: small enough to be implemented quickly, yet meaningful enough to demonstrate real value. The solution can prove itself in practice before further investment is made.