Universities

Three ways we work with schools

Most schools start with one of these and add the others as students and staff see what's possible.

• In the classroom

  We teach business and design students the levers that change what a language model does — which model, what it's told, what it can look up — so they can design with one as an ingredient in what they make. That might be a single session or several weeks inside a course.

  Workbench→Napkin→
• Tools in students' hands

  Students get their own access to the tools we use with clients, so they tailor language models and build working AI projects rather than only reading about them.

  Workbench→Caliper→Prism→
• AI for the institution

  Assistants and workflows for the school itself, like orientation, that answer from your own documents in your own language.

  Workbench→

• For three years, UC Berkeley has invited us to teach its graduate design students how to tailor language models.

  Each year we lead two to three weeks of guest sessions in Technology Design Foundations and other graduate courses, and every graduate design student gets access to ZeroWidth to build with what they've learned.

  

• At the Institute of Design, we've brought course work, tools and an orientation assistant for new students.

  Students worked in our tools as part of their courses. The orientation assistant answers questions about courses, campus policies and the first week from ID's own documents, and the questions students asked showed where the school's resources fell short, especially for international students.

  

What a class gets

• One workspace for the class

  The instructor sets it up once and the whole class works in it, so everyone starts from the same place and the term's work isn't scattered across personal accounts.

• No model accounts to set up

  Tailoring and running language models is part of the workspace, so students don't create accounts with model providers or put a card down to do the coursework.

• Work an instructor can open

  Students hand in what they built as a link. Whoever opens it doesn't need an account, so feedback doesn't wait on anyone signing up.

• Theirs after the course

  What a student builds exports as a file they keep, so the work goes with them when the semester ends.

What your IT team will ask

• Student work isn't training data

  We don't use what your students build to train models, ours or anyone else's, and we don't hand it to model providers for training.

• No surprise bill

  A class works through the inference its plan includes, then any credit the school has bought. When both are gone, runs stop until someone adds more, so there's no overage to explain at the end of term.

• Single sign-on

  On the Enterprise plan, students and staff sign in with the identity provider the university already uses.

• Choose the models

  An Enterprise workspace can limit which model providers its flows are allowed to use.

Three ways of working with us:

• Inside your team

  We join your team to find the work worth changing, build it with you, and prove it holds before it scales.

• Alongside you

  We work with your team a couple of times a week, hands-on with the technical problems and the strategic ones.

• With our tools

  Your whole team works in the tools we use on our own client work, in one shared workspace.