Jeff Witters

Meet Jeff

5.00 (1)

3 following9 followers

I design AI products that people can actually understand, steer, and trust.

Most AI features do not fail because the model is weak. They fail because the interface never made the model legible. Users cannot tell what it just did, cannot correct it when it is wrong, and cannot predict what happens next. That is a design problem, and it is the one I solve.

My background is complex, high-stakes software: a multichannel experience application for BCG, a situational awareness interface for a defense platform at BAE Systems, investor dashboards built on Tableau, a banking platform, a commercial real estate deal platform, and SaaS products including LifeBooster and Namebase. Dense data, real consequences, and users who cannot afford to be confused. AI products are that same problem with a new source of uncertainty layered on top.

What I work on:

Conversational and agentic interaction design. Prompting and input patterns, confirmation and undo, streaming and partial states, memory and context, agent hand-offs, and the failure paths most teams get to last.

Trust and control patterns. Disclosure, provenance, confidence, correction, and recovery, documented so behavior stays consistent as more of your product gets touched by a model.

Design systems that account for AI. Most systems have no answer for streaming text, confidence states, agent actions, or citations, so teams improvise and the product starts to feel unreliable even when the model is good. I build the layer that covers it.

End-to-end product design. Discovery through production-ready handoff, working directly with your engineers rather than throwing files over a wall.

Ways to start, depending on where you are: a two week AI Product UX Audit if you have shipped something and it is not landing, a three week Conversational and Agentic UX Design Sprint if the experience is still undecided, six weeks for a full MVP design, or an embedded weekly retainer if you are shipping continuously and need design judgment in the room.

I am based in Greenville, South Carolina, work US Eastern hours, and collaborate with teams across the US and Europe. If you are building something with a model at the center of it and the experience is not yet where it needs to be, get in touch.