CONTRACT CONDITIONS:
Commitment About 1.5 days a week (roughly 10–11 hours,
based on 7-hour days)
Length 6 weeks, Nov. 9 to Dec. 18, 2026
Remote or Hybrid Toronto - EST time
Leo Toronto is a creative advertising agency within the global Leo network. The agency partners with major brands to develop creative ideas and campaigns, bringing together strategy, creativity, design and production.
THE PROJECT
We're a small team embedded with a large consumer tech client, working on an AI assistant in the health and wellness space. Instead of only replying with text, the agent decides when to show interactive UI, like cards, comparisons or planners, right in the chat. In the last round we built this as a working prototype with a real model behind it.
This round is about finding out which of those tools people actually want, and what the assistant needs to earn a bigger role. Over six weeks we'll put real tools in front of real people across two rounds of testing, and turn what works into a defined system.
WHAT YOU'LL DO
- Shape the agent people talk to in testing. Write and tune the prompts, tool descriptions and behaviour rules that decide when each tool shows up.
- Design the moments that matter most: onboarding, when the agent asks for personal information, and when one assistant should hand the conversation to another.
- Map edge cases for the core tools: the golden path, ambiguous intent, and the fallback at each step.
- Observe user testing sessions and fold what we learn back into the design.
- Join daily syncs with the client. A lot of the value is in those conversations: pressure-testing ideas, flagging risks early, and helping the client make calls.
- Work closely with our UX designer and creative technologist, who build your thinking into a working prototype every Friday.
WHAT WE'RE LOOKING FOR
- You've designed conversational or agentic experiences on top of LLMs, and you've done it with clients. You walk a client through trade-offs and help them land a decision, not just hand over a flow.
- You think in edge cases. You 're the person who asks "and what if they say this?" and then writes down the answer.
- You're comfortable writing and editing prompts, tool descriptions and behaviour rules, not only sample dialogue.
- You're happy working scrappy. Work in progress over polished decks.
Why generative UI matters here. It's a paradigm that's hard for a lot of people to wrap their heads around. Part of this role is helping the client think in it, so the more you've lived it, the better.
NICE TO HAVE
- Experience with health or wellness products
- Working with Claude or other models directly, or reading prompts in a codebase
- Experience sitting in on user research and turning findings into design changes
- Multi-agent experiences, where one assistant hands off to another
HOW TO APPLY
Send us your resume and a short note that covers:
1. Start with your AI product example, in two or three sentences: what the product was, what the model did beyond chat, and what you designed.
- A link to the work: a flow, edge-case map, prompt or behaviour rules you wrote.
We'll only be able to review applications that include an example of AI product beyond a basic chat app.
