The idea sits at the intersection of two things that already exist but haven't been combined well: personalised running plans and conversational AI. Plan generator apps produce structured training schedules but the relationship is one way. A real coach adapts. They ask how you slept, how your legs feel, whether last week's long run took more out of you than expected. They adjust based on the answer. Vanta was built around the idea that an AI coach could do the same, and that the conversation itself was the product.

‍

The case for AI coaching goes beyond convenience. A real coach costs upwards of £40 an hour, requires scheduling, and assumes a level of social confidence that not everyone has. Reaching out to a stranger about your fitness, your struggles, your goals, that's an extrovert friendly interaction that a lot of people quietly avoid. AI removes that friction entirely. The conversation can happen at 6am before a run or 11pm after a hard week, on your terms, without the awkwardness.

‍

The brief was primarily UI focused. The founder, a recreational runner with a clear product vision, needed the concept brought to life visually. Recognising that the concept hadn't been stress tested strategically, I introduced a mapped assumptions exercise early in the process to give the work a stronger foundation before any design decisions were made.

Informed by informal conversations with runners, the mapped assumptions articulated the things that would need to be true for the product to work, and the rationale behind each one.

‍

Runners using plan generators find them too rigid. Training reality changes week to week based on energy, injury, life, and current apps have no real mechanism to respond to that.

‍

AI conversation is something runners would actually engage with. The success of conversational interfaces in adjacent spaces suggested the interaction model had real potential, particularly for a solo sport where external accountability is often the missing ingredient.

‍

Personality led coaching is a meaningful differentiator. Someone who wants a tough coach has a fundamentally different relationship with training than someone who wants encouragement. Letting users choose their coaching style wasn't just a feature, it was the core of what made the product feel personal.

‍

The market sits at the intersection of people already using structured training tools but finding them too passive. That's a real and reachable audience.

‍

Health data integration was identified early as a natural extension of the concept, syncing with Apple Health or Google Fit to give the AI richer context for its coaching decisions. It was deliberately parked for a later stage, the right idea for the roadmap, too much complexity for a concept build.

‍

Brand Direction The first direction explored a dark, high energy aesthetic, black, neon, bold type. It was the direction I led with when presenting initial concepts, and while it looked confident it also looked familiar, too close to the visual language of UK gym brands and fitness apps that already crowd the market. Putting our heads together it became clear it needed refinement, and that the more considered direction was the stronger call.

‍

The second direction took its cues from Apple and Google, clean, considered, typographically led, with a soft palette of blues, yellows, and greens added to keep the brand feeling grounded rather than cold. Satoshi was chosen as the typeface, simple, modern, and highly functional, a font that did the same job as the product itself. Techy without being aggressive, minimal without being sterile.

‍

A set of colour tokens and a typography scale were established early to keep the UI consistent as screens developed, giving the product a coherent visual foundation without over-engineering the system at concept stage.

The interaction model was built around the AI coming to you rather than something you had to open and navigate. The coach checks in, asks how you're feeling, responds to what you tell it, and adjusts accordingly. It was designed to feel less like an app and more like a running partner who happened to be available whenever you needed them.

‍

The onboarding flow established that conversational tone from the first screen. Rather than a standard form, users moved through a series of selections, training goal, race date, current fitness level, available days, injuries or concerns, building a profile through choices that felt like the start of a coaching conversation rather than a data capture exercise.

‍

The final onboarding step was coach selection: Tough, Nerd, Zen, or Supportive. Each represented a distinct AI personality and coaching style, giving users immediate ownership over the kind of relationship they were entering into. The visual language carried through every screen, with the soft palette and clean typography keeping the experience feeling calm and considered throughout.

The brand and UI gave the founder a conversation ready set of visuals to take into early discussions. Informal conversations with potential backers responded positively to the concept, validating the core direction. The next step would be a proper research and testing phase before anything was formally put on the table.Takeaways Introducing assumption mapping on a UI led brief was the right call. It's a lightweight intervention that adds real value when the brief doesn't naturally include a research or strategy phase, and something worth doing as standard on early stage concept work going forward.