OmniCastAI.
My Role
Senior Product Designer
Period
2-4 weeks
AI-powered publishing assistant that predicts performance and optimizes multi-channel campaigns in real-time.
"AI-powered publishing assistant built on 8 interviews and 4 competitor audits. Projected 30–40% time savings and 15–20% engagement uplift across 6 platforms."
The Challenge
Agencies, marketers, and creators today juggle multiple social platforms with entirely fragmented workflows. Every campaign involves switching tools, guessing optimal posting times, manually writing captions, and waiting until after publishing to see what worked.
This inefficiency leads to lower engagement, wasted ad spend, and slower turnaround — hurting both campaign ROI and brand growth.
The core opportunity wasn't just a faster publishing tool. It was designing an AI-powered assistant that could predict the best way to publish before content goes live, continuously optimise campaigns in real time, and reduce repetitive work without taking away user control.
Research
I ran interviews with 5 marketers and 3 agency owners, a competitor analysis of Meta Business Suite, Hootsuite, Later, and Buffer, and workflow mapping of typical campaign planning cycles.
What I found:
- Users wanted AI help, but not "black box" automation — they needed to see the reasoning
- Predictive analytics mattered far more than retrospective reports
- List view and calendar view were equally important for power users
- Most AI integrations in existing tools are add-ons bolted on the side, not core to the workflow
Design Principles
Transparency — Always show why AI suggests something.
Actionability — Every suggestion leads to a clear next step.
Seamlessness — AI embedded in every step, not hidden behind a separate tab.
Concepts Explored
Dedicated AI tab for all recommendations — Rejected. Adding a separate tab created friction by requiring users to leave their active workflow.
Fully automated publishing without user review — Rejected. Removing the human in the loop reduced trust and control.
In-line AI suggestions integrated into dashboard, post creation, and calendar — Chosen. This matched existing mental models and built trust incrementally rather than demanding it upfront.
The Solution — OmniCast AI
OmniCast AI is an AI-first publishing workflow where every major action is guided by predictive intelligence.
AI Insights Panel
Surfaces 3 priority actions at login — giving users immediate performance gains without hunting for insights buried in dashboards.
AI Content Creation
Generates tone-specific captions and hashtags on demand — reducing content creation time and eliminating the blank-page problem for high-volume publishers.
Predictive Scheduling
Suggests optimal posting windows with a percentage engagement uplift estimate — so users know not just when to post, but why that window performs.
AI Content Calendar
Month and list view with live re-scoring as campaigns run — campaigns stay optimised rather than set-and-forgotten.
Continuous Optimisation
Monitors live posts and flags underperforming content before the window closes — reducing wasted spend on posts that need to be pulled or boosted.
Design System
Two UI modes for two contexts:
- Liquid Glass (Apple-inspired) — for creative showcase and presentation
- Clean Professional — for daily operational use
Choosing fewer but deeper AI features over many shallow ones kept the interface focused. Collaboration with a marketing lead shaped how AI confidence scores were displayed — initially hidden, they were made visible after feedback revealed that transparency was the single biggest trust driver.
Outcome
- Estimated +15–20% engagement uplift based on AI recommendations in test scenarios
- 30–40% time saved on repetitive publishing tasks
- Clearer performance tracking led to better campaign prioritisation across teams
What I Learned
AI adoption depends heavily on explainability. Users trust AI more when they can override it without friction — the ability to ignore a suggestion matters as much as the suggestion itself.
Predictive insights proved more valuable than historical analytics in fast-moving content cycles. Knowing what will perform beats knowing what already didn't.
The framework is designed to expand: AI-driven ad targeting, creative testing, and competitor tracking are natural next layers on the same foundation.
AI embedded in every step, not hidden in a separate tab — that's the difference between a tool users trust and one they work around.