![\"\"](\"https:\/\/nexpert.fi\/wp-content\/uploads\/2026\/05\/CASE-RD-in-Public-sector-.jpg\" "\\"\\"")
<\/figure>\n\n\n\nThis made feedback valuable, but difficult to manage. Although individual feedback might have been a small observation, as a whole, it concerned the organization's ability to learn from its own operations, projects, stakeholders, and the daily work of experts. The work identified the same starting point: feedback was multi-channel and continuous, and scattered information caused rework, searching, and delays in expert work. <\/p>\n\n\n\n
Why was this made into an R&D project? <\/h2>\n\n\n\n
Feedback management could have been attempted by creating a new guideline. However, in this case, that would not have been enough, because the topic was too broad and too tied to the daily work of experts. What was needed was a knowledge base, practical experiments, and above all, the involvement of the organization's own experts. If the goal is a learning organization, the model cannot be built solely from an external perspective. Those doing the work know best where information truly flows, where it breaks, and what kind of solutions could work in daily life. <\/p>\n\n\n\n
The TKI project offered a suitable structure for this. The project had a clear scope, a learning organization, and a limited timeframe. At the same time, the objective was left sufficiently broad to allow room for iterative development. This was important because, at the outset, it was not yet known which individual solution would be the most impactful. <\/p>\n\n\n\n
Additionally, the perspective of automation and artificial intelligence was desired from the outset of the project. Process development can no longer be based solely on people manually entering, copying, and compiling information. If information is scattered, in different formats, and with different people, the possibilities of technology must be included during the development phase itself, not just at the end.\u00a0<\/p>\n\n\n\n
Expert's role: structure, facilitation, and interpretation <\/h2>\n\n\n\n
Nexpertin's role was to plan and advance the RDI project using a research-based development approach. The project's progression was planned independently, but each phase was approved by a steering group. The steering group provided direction, commented on progress, and made decisions about what to focus on next. <\/p>\n\n\n\n
Work progressed in PDSA cycles. First, a light improvement measure was planned, then it was implemented and data was collected. After that, the findings were analyzed, and based on them, the next development round was planned. Finally, the results and the proposal for the next round were presented to the steering group for decision-making. <\/p>\n\n\n\n
In practice, the work involved planning and facilitating workshops, conducting surveys, analyzing data, conducting interviews, making observations, testing AI tools, and compiling results to support decision-making. The work also had an important interface with technology development. Nexpert participated in the AI competence center's development sprints, tested tools under development, and contributed a subject matter perspective from the design industry to the technical development. <\/p>\n\n\n\n
This interpreter\u2019s role was crucial because artificial intelligence solutions do not develop in isolation from everyday work. They need to understand concepts, ways of working, responsibilities, and the points where an expert\u2019s work is actually burdened. In addition to research, Nexpert\u2019s task was to structure and communicate the infrastructure sector\u2019s substance needs to technology developers, as well as to test AI features transitioning into production. <\/p>\n\n\n\n
The worker knows best how the work should be done <\/h2>\n\n\n\n
In the project, experts were involved using several methods. For example, Learning Cafe and Me-We-Us methods were used in workshops. Additionally, data was collected via a Microsoft Forms survey and Teams Polls. Lean tools were utilized in the analysis, such as A3 problem-solving, Ishikawa diagrams, and the 5 Whys method. The work was supplemented by observations during workshops, meetings, and artificial intelligence development sprints. <\/p>\n\n\n\n
No single \u201cbest\u201d method was chosen in advance. The findings from the previous round influenced what was done next. This was the whole idea of the project: not to lock in a solution too early, but to let the data, experts, and observations guide the next step. <\/p>\n\n\n\n
One particularly useful practical observation related to Teams Polls. In large expert meetings, communication easily becomes one-sided. Especially in official duties, meetings often deal with new regulations, requirements, and operating procedures that are important to internalize. Teams Polls offered a lightweight way to make meetings more participatory. It provided a quick way to get the perspective of a large group of experts on what is important, what to focus on, and whether the direction is correct. At the same time, participants stayed more engaged. This did not require a separate, heavy workshop, but rather the inclusion of one well-thought-out question as part of a normal meeting.\u00a0<\/p>\n\n\n\n
From a mixed initial assessment to a clear overall picture <\/h2>\n\n\n\n
One of the important outcomes of the project was the structuring of the feedback field. Initially, the current state was described as a broad and difficult-to-grasp mind map. During the work, it evolved into a clear overall picture resembling an Ishikawa or fishbone diagram. This made it possible to see at a glance what the main themes of the feedback were and what kinds of channels belonged under them.<\/p>\n\n\n\n When the whole picture becomes visible, it can be discussed collectively. No longer are we just talking about individual feedback or isolated observations, but we see how feedback relates to practices, people, data, systems, metrics, and the operating environment. <\/p>\n\n\n\n This kind of structure is especially needed in large organizations where information related to the same entity can come from multiple directions. Without a visible structure, the entity easily becomes dependent on the people who happen to remember where things were processed. <\/p>\n\n\n\n The project also examined the possibilities of artificial intelligence in processing feedback and supporting expert work. AI was tested, among other things, in material classification, creating meeting minutes, data analysis, and producing presentation materials. A local AI service was utilized in the work, and the functionality of different language models was also tested in limited use cases. <\/p>\n\n\n\n An important observation was that AI alone does not solve feedback management. First, one must understand the process, feedback channels, responsibilities, and decision-making points. Only then can it be assessed where AI can be beneficial. In some tasks, AI clearly sped up analysis. In others, it still required a lot of checking and correction. For example, in the production of presentation materials, it was observed that AI could omit parts of the material or make errors related to text layout and formatting. This reinforced a practical principle: AI should be used where its benefits are clear, but expert evaluation cannot be bypassed. <\/p>\n\n\n\n One of the most interesting lessons from the project related to the utilization of the SPC i-card in management. <\/p>\n\n\n\n The SPC control chart, or management chart, makes process behavior visible through data. Instead of management intervening with individual actions or random observations, the chart shows when the process variation deviates from normal. This allows attention to be focused on the areas where the process truly needs investigation. <\/p>\n\n\n\n In the project, the SPC i-chart was utilized to review the progress of the R&D project and the lead time of technical development. The measurement showed, for example, points where process stability weakened, as well as natural variations, such as the impact of the summer holiday season on progress. In the work, the SPC i-chart is described as a diagram that visualizes the statistical behavior of a process, which can be used to monitor the improvement of the process flow and highlight observations that require closer examination.<\/p>\n\n\n\n This is important from a management perspective. Once measurement points and limits are agreed upon, you don't need to intervene in everything. Attention is focused on deviations, not on monitoring every single task. When used well, a metric can reduce micromanagement and bring more facts to the discussion. <\/p>\n\n\n\n This observation is particularly useful in large projects and service provider collaborations. When tasks are managed using data rather than just a gut feeling, it's possible to more quickly see where the process is actually getting stuck.<\/p>\n\n\n\n The project resulted in a description of feedback management methods, concrete action proposals, and a five-year roadmap for developing a learning organization. The thesis abstract mentions the results as a description of feedback management methods, recommendations for experiments in the following year, and a five-year roadmap for developing a learning organization.<\/p>\n\n\n\n However, what was created alongside the main product was equally important. <\/p>\n\n\n\n The project also generated understanding of other development entities. At the same time, the organization moved forward with quality management of design related entities, such as developing self-assignment and describing the quality management process. The feedback management R&D project helped tie these matters together into a bigger picture: how feedback, quality management, information flow, artificial intelligence, and learning are interconnected. <\/p>\n\n\n\n This is often the value of an R&D project. When a project is clearly defined but implemented openly enough, it doesn't just produce one report. It also creates new common language, a better situational picture, and insights that can be utilized in other projects. <\/p>\n\n\n\n If an organization receives a lot of feedback, development ideas, and observations, the first question is not \u201cwhat system should these be recorded in.\u201d. <\/p>\n\n\n\n A better question is: how does learning come from this?<\/p>\n\n\n\n Feedback management is not just about collecting information. It's about the ability to structure information, identify recurring themes, choose what to focus on, and translate findings into decision-making. In large expert organizations, this requires structure, but also space for the experts' own observations. <\/p>\n\n\n\n The key aspect of this project was that the development was not based on a ready-made model brought into the organization from the outside. First, the current state was determined, then experts were involved, the findings were analyzed, and the next step was built based on them. <\/p>\n\n\n\n At the same time, the role of artificial intelligence was put into perspective. AI can assist with classification, analysis, summaries, and data processing. But if the process is unclear, AI will simply automate the ambiguity. Therefore, understanding the process, structuring feedback channels, and clarifying decision-making must be done first. <\/p>\n\n\n\n Nexpert helps expert organizations implement development in situations where information, processes, people, and technology are intertwined. <\/p>\n\n\n\n This case involved an R&D project that combined service design, Lean methodologies (systemicity), AI utilization, expert engagement, and decision-support reporting. The end result was a structure that can transform fragmented feedback into learning and continuous improvement. <\/p>\n\n\n\n A similar approach is suitable for organizations where a lot of development work is done, but the learning too easily remains within projects, meetings, systems, or in the memories of individuals. <\/p>\n\n\n\n When information is made visible, it can be managed. When experts are involved, solutions become closer to everyday life. When measurement and artificial intelligence are introduced at the right point, development does not remain just talk.<\/p>\n\n\n\n![\"\"](\"https:\/\/nexpert.fi\/wp-content\/uploads\/2026\/05\/Sekavasta-alkukartoituksesta-selkeaksi-kokonaiskuvaksi.png\" "\\"\\"")
<\/figure>\n\n\n\nArtificial intelligence through understanding the process <\/h2>\n\n\n\n
Measuring reduces micromanagement <\/h2>\n\n\n\n
![\"\"](\"https:\/\/nexpert.fi\/wp-content\/uploads\/2026\/05\/Mittaaminen-vahentaa-mikromanageerausta.jpg\" "\\"\\"")
<\/figure>\n\n\n\nEnd result: method description, roadmap, and broader understanding <\/h2>\n\n\n\n
![\"\"](\"https:\/\/nexpert.fi\/wp-content\/uploads\/2026\/05\/tiekartta-nexpert.png\" "\\"\\"")
<\/figure>\n\n\n\nWhat can a corresponding organization learn from this? <\/h2>\n\n\n\n
Nexpert's perspective\u00a0<\/h2>\n\n\n\n