Generating creative inside Mandala - the AI reads the marketing brief and the selected audiences, then produces on-brand visuals tuned to each segment.
WhyI wanted the marketer to never leave their canvas. If they've already loaded a brief and picked their audiences, generation should just happen there - not in a separate tool that asks them to re-explain themselves.
Drafting a full marketing plan with AI - objectives, channels, audiences, and budget shaped from the brief, then handed to the marketer to edit and approve.
WhyA blank-page marketing plan is the part nobody wants to start. I framed AI as the first draft, not the final word - it does the structuring, you do the judgment. The marketer stays in control because every field is editable the moment it lands.
Asset Studio - a dedicated workspace for generating, remixing, and resizing creative variants across formats, all from a single source brief.
WhyCreative teams were exporting the same asset eight times for eight placements. I designed Asset Studio around one source of truth - generate once, remix and resize from there - so the brief stays intact across every variant.
Editing a generated image directly inside Mandala - swapping elements, adjusting composition, and re-rolling regions without leaving the canvas.
WhyGeneration always felt like the easy part - the hard part is the next ten tweaks. I kept editing on the same canvas as generation so marketers can iterate in place instead of round-tripping into Photoshop.
Generating audiences from a plain-language prompt - the AI translates intent into segments, sized and ready to plug into a campaign.
WhyAudience builders historically punished people for not speaking the tool's language. I leaned into plain language because that's how marketers actually describe who they're trying to reach - the AI translates intent into a real, sized segment.
Bring-your-own-data audience builder - upload first-party signals, blend them with Google's data, and shape a custom audience inside the same workflow.
WhyFirst-party data is a marketer's most valuable input, but most tools treat it like a separate world. I designed BYOD to feel like one continuous workflow - your data, Google's data, blended in the same surface, so the boundary stops mattering.
A spatial view of an audience - clustering interests, behaviors, and overlaps so marketers can see who's actually inside the segment before they spend against it.
WhyAn audience is usually a number on a card - I never felt like that was enough to spend money against. I pushed for a spatial view so marketers can actually see the shape of a segment before they commit budget to it.
Generating audiences right inside Mandala - the marketer describes who they're trying to reach and the canvas spins up segments alongside the creative.
WhyMarketing decisions are interlocked - audience shapes creative, creative shapes brief - but tools insist on isolating them. I put all three on one canvas so changing one nudges the others, which is closer to how the work actually feels.
Live campaign performance with AI suggestions sitting next to it - what's pacing well, what's underperforming, and the next move the marketer should consider.
WhyDashboards tell you what happened; marketers need to know what to do next. I paired the live metrics with AI suggestions sitting right beside them so the insight and the next move share the same glance.
Upload an existing marketing brief and the AI extracts the goals, audiences, KPIs, and constraints into structured fields - ready to drive the rest of the workflow.
WhyEvery marketer I talked to already had a brief in a doc, deck, or PDF - and was retyping it into our tool. I designed the upload-and-extract flow to respect that work: keep the source, lift the structure, let them edit anything the AI got wrong.
An intelligent launch flow - the system pre-checks the campaign, flags risks, suggests fixes, and walks the marketer through the final steps before going live.
WhyLaunches are where confidence wobbles - one missed setting and a campaign underperforms for weeks. I treated launch as a guided moment, not a button: pre-checks, plain-language risk flags, and clear fixes so the marketer goes live knowing what they shipped.
Direct-manipulation editing inside Mandala - drag elements on the canvas to reframe, resize, and re-compose generated imagery without touching a panel of controls.
WhyPanels of sliders and dropdowns slow creative work down. I wanted the canvas itself to be the control surface - drag to reframe, pull to resize - because that's how designers think about composition in every other tool they use.
Behind the pixels
A look at the rough work along the way.
Swipe to browse - tap to expand.