--- title: "Instill AI Document Intelligence Design Case Study | Donux" description: "How Donux designed Instill AI, an AI workspace that turns documents into structured, cited tables, from onboarding to the marketing website." url: "https://donux.com/case-studies/instill-ai" language: "en" --- ![Instill AI logo](https://images.app.donux.com/unsafe/fit-in/240x0/case-studies/instill-ai/logo.png) # How Instill AI moved document work beyond the chat box ![Instill AI onboarding step to choose a goal](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/hero-01.png) ![Instill AI website hero with a document review prompt](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/hero-02.png) ![Instill AI collection setup with chat prompt and templates](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/hero-03.png) ![Instill AI project view with agent assistant panel](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/hero-04.png) ![Instill AI collection table with AI autofill column settings](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/hero-05.png) Industry AI & Machine Learning Services Product Design, UX Design, Website Design, Brand Identity 1/ ## The situation **Instill AI** is an AI workspace for complex documents: contracts, reports, data rooms and research. Most professionals who decide from documents rely on generic AI chat. Answers get lost in chat logs, sources are hard to trace, and with many documents it's unclear where each point came from. Instill AI needed an experience that turns document work into **reusable, verifiable context** instead of one-off conversations. ![Chat answers scattered across multiple documents and file types](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/image-01.png) 2/ ## Build context once, reuse it Collections became the heart of the workspace: tables where AI extracts and cross-references information from every document. Work builds up over time and can be sorted, filtered and exported, instead of disappearing into chat history. ![Instill AI collection table with AI autofill and context settings](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/image-03.png) ![Instill AI collection templates, uploaded files and collection overview](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/image-05.png) 3/ ## Answers you can verify Every extracted value links back to its page and paragraph, and the agent summarizes key terms and risks with citations. Trust in AI output comes from **being able to check it**. ![Instill AI lease review table with page references and cited summary](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/image-09.png) 4/ ## Value from the first session A goal-based onboarding and ready-made templates let new users start from a use case, from scratch or by dropping files in, with the agent always available on the side. ![Instill AI onboarding asking users what they plan to do](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/image-02.png) ![Instill AI empty project with options to create a collection](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/image-04.png) 5/ ## One story, from website to product We extended the design language to the marketing website and pricing pages. ![Instill AI website homepage and pricing on desktop and mobile](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/image-10.png) They explain the difference between chatting with documents and building context, in the same words users meet inside the product. ![Instill AI logo mark](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/image-06.png) ![Instill AI workspace tree with projects and collections](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/image-07.png) ![Bar chart illustration from the Instill AI visual language](https://images.app.donux.com/unsafe/fit-in/1440x0/case-studies/instill-ai/image-08.png) ## Is your AI product stuck at "chat with your documents"? We design the experience that comes after the chat box: structured, verifiable and built to be reused. [Let's talk](https://donux.com/contact-us) [Explore Product Design](https://donux.com/services/product-design) ## More Case Studies [![Bonobo AI inbox with an AI-generated proposal summary](https://images.app.donux.com/unsafe/case-studies/bonobo/hero-01.png)](https://donux.com/case-studies/bonobo) AI & Machine Learning ### How Bonobo AI turned a quote generator into an AI-powered quoting platform [Explore the project](https://donux.com/case-studies/bonobo) [![Mavexa AI event planning workspace](https://images.app.donux.com/unsafe/case-studies/mavexa/CtgNa3i1fX59P9vslHwLkw33A.png)](https://donux.com/case-studies/mavexa) AI & Machine Learning ### How Mavexa made B2B event planning as simple as a single prompt [Explore the project](https://donux.com/case-studies/mavexa) [![Blue.ai platform interface](https://images.app.donux.com/unsafe/case-studies/blue-ai/uzXYj3jBxT9m9q2gUnbKQpsu9M.png)](https://donux.com/case-studies/blue-ai) Marketing & Sales ### How Blue.ai transformed their omnichannel CX platform in 12 weeks [Explore the project](https://donux.com/case-studies/blue-ai)