Dobbin’s brand foundation started as a 52-page Google Doc that was hard to read and scan, and some customers had never documented their business at brand level before.

I redesigned it as a living document: sections made up of content cards that are quick to scan, so more customers found it easier to browse and understand. Teams can edit any card, track how each section is used and see who last worked on it. Dobbin also tracks which parts of the Foundation shaped its answers. In review mode, customers apply or dismiss its suggested changes.

Every company runs on documents it relies on: strategy docs, contracts, product lists and more.

The Knowledge Base brings them together in Dobbin HQ. Each document can be global, across every Slack channel, or kept to specific channels, so departments keep their privacy. HQ shows how often each document is used, and when a conversation is missing context, Dobbin writes a new “learned” document to fill the gap.

Much of a company’s knowledge lives in the tools it already uses.

Customers can connect them to Dobbin through integrations, which work just like documents and give Dobbin the company’s real data to answer with.

Trust drops fast when an AI answer misses and teams can’t see why.

I explored a canvas where teams unpack prompts Dobbin didn’t fully solve: the knowledge each answer used, or was missing, and the gap behind it. It’s multiplayer, so teammates can fix gaps together, and the nature of the canvas means users carry on exploring and refining their system.

Throughout the experience, whether that’s onboarding, Slack, Claude or HQ, I always tried to explain how Dobbin works and show its status. Small cards, like team usage in HQ, show customers what Dobbin is doing for them. This again helps build trust, especially with customers who are less technical.

Shipping weekly with a team of two, I needed to move fast and still have something robust to build on.

I built Dobbin HQ’s design system in Figma while designing core parts of the UX, then took it into code with Claude. Tailwind is the underlying structure, with Dobbin’s colours, fonts and UI on top. Claude and Figma stay in sync through the Figma Bridge MCP, and the design system also lives in the repo, so developers share the same source of truth.

Dobbin needed personality in a crowded AI market: a lean brand, just for the product, that we could build on later.

I designed it in a short, dedicated sprint. The Dobbin D is built from two elements: a solid shape for the foundations of the knowledge system, and an overall form that feels conversational.

Marketing needed a steady stream of on-brand posts, without a designer in every loop.

I built Dobbin Studio, an internal tool where anyone on the team can create and test LinkedIn posts. I designed the templates in Figma and synced them to Claude through the Figma MCP, so the look stays in design’s hands while Studio’s constraints keep everyone on brand. Dobbin is connected too, so you can check your copy and ideas against your brand foundation.

We already use it internally, and we’re adding more options and the ability to create simple videos.

Early traction with a two-person team: 8 of our first 10 customers are still with us.

• Since the Foundation and Knowledge Base redesign, customer success no longer needs to help customers with either.
• Our first enterprise client, alongside Area 17, Mark Seliger Studio and Evite.
• Brand, marketing site, Dobbin HQ and its design system, designed and shipped in 3 months.

“There’s something really different happening with what you built.”A Dobbin customer