Research groups are experimenting with large language models, but most use stays at the level of one-off questions in a chat window, with the context rebuilt every time. For Hubrecht institute in the Netherlands, FutureWater is delivering a tailored AI training for around twenty researchers. Each participant brings one task from their own work and keeps it through the morning, building it up in three rounds: a project workspace holding their own material, a reusable skill that captures how they work, and connections to the systems and sources they use daily. The aim is that everyone leaves with one working setup of their own.
Most researchers use a large language model as a better search engine: ask a question, get an answer that is just about acceptable, then spend a considerable amount of time editing it up to standard. The next day a similar problem arises and the process starts from scratch. What changes this is not a better prompt, but giving the model a different job: not answering questions about your work, but carrying out a defined part of it. That happens in your own folders, on your own data, within rules you set and checks you can verify yourself, that run every time. That is the difference between an answer and a result you can use in your workflow.
The Kops Research group, part of Hubrecht Institute, asked FutureWater to give the researchers in one of its groups a practical grounding in working this way. The training is built around a half day workshop for around twenty participants, held during the group’s retreat. This retreat was wholly centered around optimal and responsible use of AI and included a preceding talks by the research institute of about using AI within their research projects. The training by FutureWater expanded on this in a practical session in which each participant picks a task from their own work.
The workshop runs that one case through three rounds, each with a short explanation, a long hands-on block and a joint review. The first round sets up a project workspace and structure holding the all necessary files and standing instructions. The second is aimed to capture (a step of) their method as a reusable skill, so the same work runs the same way when it is reused by a colleague. The third adds connections to integrate the solution in their own systems, and subagents that pick up parts of a task in parallel. FutureWater demonstrates each round with tools it has built for our own work.
Alongside the rounds, the programme covers choosing between AI tools, prompting, data analysis, adhering to figure conventions , literature screening with source verification, scientific writing for non native speakers, and building in manual checks. Slides and training material go to all participants afterwards, and remote support is available.
