Jan Bosch is beginning to see the emergence of a new category of AI startups: Instead of trying to automate creative work, they target the mundane administrative overhead that slows organizations down.
Administrative work is the hidden tax of modern organizations. Companies spend enormous amounts of time filling out questionnaires, answering requests for documentation, resetting passwords, approving access rights and routing internal tickets. These tasks are rarely strategic, but they are essential.
For years, companies have tried to reduce administrative overhead through robotic process automation (RPA). Traditional RPA tools mimic human actions in software systems: clicking buttons, copying fields between applications and executing predefined workflows. This works well when processes are stable and structured, but it breaks down quickly when documents vary, systems change or decisions require interpretation. Maintaining RPA scripts often becomes a project in itself.
Generative AI changes the equation because it can understand unstructured information, such as emails, PDFs, questionnaires and policies, and reason about their content. Instead of automating the mechanics of a process, GenAI can automate the cognitive work inside the process: interpreting requests, finding relevant information and drafting responses. As a result, while RPA automated repetitive actions, GenAI increasingly automates repetitive thinking, opening the door to eliminating large portions of administrative work altogether.
A new generation of startups is discovering that large language models and workflow automation are particularly well-suited for exactly this kind of work: document-heavy, repetitive and rule-driven processes. Instead of trying to automate creative work, these companies target the mundane administrative overhead that slows organizations down. Two particularly good illustrations of this trend are Sunhat and Moveworks.
Consider a typical B2B supplier relationship today. Before doing business, customers increasingly require suppliers to complete extensive questionnaires. These might cover sustainability (for example Ecovadis), compliance with ISO standards, cybersecurity practices or ESG reporting. The problem is that every customer asks slightly different questions. Sustainability teams and compliance teams spend days or weeks collecting answers from internal experts, locating evidence in documents and filling out yet another form.
Sunhat addresses this problem directly. The company provides a platform that collects sustainability data, policies and documentation in a central knowledge base and uses AI to automatically generate answers to incoming questionnaires. Instead of starting from scratch every time a customer sends a request, it identifies relevant previous answers, maps them to the new questions and generates draft responses supported by evidence from the organization’s documentation. This isn’t GenAI-only, but Sunhat promises guaranteed evidence of the data and answers provided.
The impact can be significant. What previously required weeks of manual work, including tracking down information across systems like Sharepoint, SAP or Excel, can be reduced to hours because the AI already knows where to find the relevant ‘proof.’ In other words, Sunhat turns the organization’s fragmented knowledge into a continuously reusable repository of answers. Each questionnaire improves the system, making the next one easier.
One challenge is, of course, that as with all systems, it’s garbage in, garbage out. Companies need to provide the data to the system in a timely and sufficiently structured way to ensure high-quality output. Although we all love a silver bullet, there are few of those around. Still, the broader insight is that as regulation and reporting requirements increase, companies face a growing administrative burden. Startups like Sunhat are transforming that burden into a data-driven automation opportunity.
Whereas Sunhat addresses external administrative work, Moveworks focuses on the bureaucracy inside organizations. In a typical big company, thousands of employees generate an endless stream of internal requests: “Can you reset my password?”, “How do I get access to this system?”, “Where do I find the HR policy?”, “Please approve this access request.” These usually end up in ticketing systems and service desks. Even in well-run companies, IT and HR teams spend large amounts of time answering routine questions.
Moveworks approaches this problem by creating an AI assistant that sits on top of enterprise systems and automatically resolves employee requests. The platform integrates with tools such as Slack, Microsoft Teams, Servicenow and internal knowledge bases. Employees simply ask a question in natural language and the system interprets the request, retrieves the necessary information and executes the required workflow. For example, the AI might automatically reset a password, grant access to software or route a request to the right team without human intervention. The goal isn’t to replace experts, but to eliminate the routine tasks that consume their time. Instead of manually handling hundreds of tickets, support teams can focus on the complex problems that actually require expertise.
What unites companies like Sunhat and Moveworks is a simple insight: Organizations generate enormous amounts of administrative work that adds little strategic value. Historically, this work couldn’t easily be automated because it requires reading documents, interpreting requests and combining information from multiple systems. That’s precisely the type of work where modern AI excels.
As a result, we’re beginning to see the emergence of a new category of startups whose goal is simple but powerful: Remove the friction of organizational bureaucracy. Instead of making people more productive at administrative tasks, these companies aim to make those tasks disappear entirely. To end with Peter Drucker: “The greatest waste in business is unnecessary work.”
