Handraise
Turning earned media into owned reach
TL;DR
PR teams earn media coverage but lack control over distribution, amplification, and measurement. I led design for Handraise's 0–1 AI product to solve this.
In 6 weeks, we built, tested, and pivoted. Early prototypes showed audience targeting mattered more than content generation, which reshaped the product. I redesigned the experience around an audience-first workflow, created a lightweight design system, and left the team with a validated direction and a shippable foundation.
Problem
PR has a distribution problem, not a content problem. PR teams are good at earning coverage. The breakdown happens after the story runs.
The coverage exists, but the right audiences often don't see it. Reaching them depends on marketing tools, budget, and coordination that moves slower than the news cycle. By the time a campaign launches, the moment has passed.
At the same time, expectations changed. PR teams are now asked to show measurable impact, not just impressions. But they don't have the tools to drive that themselves. The expectations evolved. The tools didn't.
The original bet
What we started with
The initial idea was straightforward: an AI tool that turns a press article into ready-to-use social posts and ads. Fast, simple, and clearly useful. I built a high-fidelity prototype and tested it with PR professionals across multiple sessions.
It worked. People could see themselves using it. But in every session, the conversation shifted to a different question.
What the testing revealed
Across 5 of 6 sessions, users moved quickly past the generated content and asked: who is this going to? How do I reach the right people?
Content wasn't the issue. They trusted Handraise to generate social media ads from articles. What they lacked was a way to identify the right audience, reach them without marketing, and know if it worked.
What we rebuilt
I reframed the product from a content generator to a distribution system. The new framing gave Handraise a clear reason to exist in a crowded field of AI-generated creative tools. Generating posts is a feature. Getting the right content to the right audience, without relying on marketing, is the product.
| Original pitch | Reframed value proposition |
|---|---|
| Input: Press article | Input: Earned media |
| Output: Auto-create social posts | Output: Article distribution to precise audiences |
| Value: Promote press coverage | Value: Unlocks additional reach |
What I designed
The audience-first workflow
If audience targeting is the core value, it has to come first — not as a step after content is created. Early versions placed targeting at the end. Users rushed through it or skipped it. By then, key creative decisions were already locked in.
Audience before content
I moved audience definition to step one. The first question is simple: who are you trying to reach? That answer shapes content, channels, and success metrics. When audience came first, users made better decisions and felt more confident in their campaigns.
AI as infrastructure, not interface
Most AI-generated social posts tools center on prompts and outputs. That model doesn't fit PR teams. They don't think in prompts and shouldn't have to.
No visible prompt box
I treated AI as infrastructure — it powers the experience but stays out of the way. The UI shows outcomes, not mechanics. If users feel like they're prompting AI too much, we've lost. It should feel like a fast, intelligent distribution tool.
Selectable post options
Content appears as selectable post options — usually three variations per audience, labeled by approach like "Recommended," "Fun," or "Educational." Users choose a direction, then edit if needed. No blank states. We avoided explaining how the AI works. Trust is earned through relevance, not explanation.
HighFive — the design system built for speed
Six weeks is short. To keep design and engineering moving together through multiple pivots, I built a lightweight design system — HighFive — in 1.5 weeks on top of shad/cn. It wasn't a full component library. It was the minimum system we needed to move fast without creating inconsistency.
- Spacing scale: 4px base with 8 steps — no guessing on layout
- Type system: 3 weights, 4 sizes — enough for hierarchy without overhead
- Color tokens: 6 semantic tokens mapped to fixed values — no hardcoded colors
- Core components: buttons, inputs, cards, segmented controls, badges, modals, loading states
- Data states: empty, loading, and error patterns defined up front
Without it, a fast-moving sprint would have produced a fragmented UI. With it, engineering could build without constant design input, and I could iterate without breaking consistency.
Prototype-led validation
Given the timeline, we validated through shipping and observing real usage rather than formal user testing.
Iteration 1 — original concept
Shipped an AI content generation flow. Observed: users engaged with output but quickly shifted focus to audience and distribution. Decision: pivot to audience-first.
Iteration 2 — audience-first flow
Shipped a reordered workflow with audience upfront. Observed: better engagement, but friction in building audiences. Too many configuration options, unclear starting point. Decision: 1-click segments and curated suggestions based on article context.
Iteration 3 — Lightweight reporting
Shipped simple funnel tracking for social post conversions. Saw a small lift in click-through rate. Creative quality drove most of the performance, with platform-specific formatting also playing a role. Decision: narrow the pilot to fewer platforms, focusing on LinkedIn and Twitter.
Key decisions & tradeoffs
Reframed the problem before designing
The brief was "AI campaign tool for PR." I reframed it to "Earned media distribution" based on what we saw in usage. Tradeoff: reset early assumptions and rebuilt the core flow mid-sprint. Outcome: aligned the team around a more valuable problem.
Audience-first over creative-first
Moved audience selection to the start of the flow, which goes against most campaign tools. Tradeoff: less immediate "wow" since content doesn't appear first. Outcome: better targeting decisions and higher confidence in the final output.
AI as invisible infrastructure
Swapped prompt boxes for simple “Generate” buttons. Tradeoff: less obvious differentiation in demos. Outcome: infinite generation, higher trust in use — users focused on outcomes, not how to prompt.
Narrow MVP scope
Excluded analytics, integrations, and collaboration to stay focused on the core loop. Tradeoff: gaps in functionality and some unmet expectations. Outcome: clearer signal on what mattered most versus what users said they wanted.
Outcomes
Validated
| Audience-first flow | 5 of 6 users completed the core loop without prompting |
| Intent to use | 6 of 6 users said they'd use Handraise over their current tool |
| Core loop speed | Article → audience → distribution completed in under 3 minutes |
Every other tool makes me do all the thinking. This one starts with who I'm trying to reach. That's the right place to begin. — Public Relations Manager, Outdoorsy