Reimagining NYX.today

Prior to my redesign, NYX.today possessed highly advanced proprietary AI capabilities, yet suffered from a fractured user experience. In the fast-paced Indian adtech ecosystem, software reliability directly correlates with brand authority. The original system relied on disconnected tools that forced users to manually jump between independent image, video, and campaign screens. This fragmented layout caused severe cognitive friction for pilot users.

This platform friction carried an immediate operational cost: product demonstrations with marquee Indian digital marketing agencies frequently stalled because the interface failed to demonstrate the true automated power of the product. This friction placed potential business pipelines worth lakhs in enterprise subscription revenue at risk due to perceived product instability.

Midway through my seven-month timeline, the core product model underwent a massive evolution. The product moved away from isolated, click-heavy execution modules toward a centralized, generative AI workspace powered by three core engines: Neo (the primary conversational orchestration hub), Pixieo (a unified image and video generation core), and Xeno (a single-prompt fully autonomous campaign engine).

During our pilot phase, the product management team realized that users increasingly expected generative AI systems to behave conversationally rather than through traditional, rigid navigation structures. Forcing a user to configure an ad campaign through endless form fields felt outdated compared to modern conversational tools. This created immense pressure to abandon our original modular homepage layout and design a unified, chat-driven workspace from the ground up.

As the solitary UI/UX designer within the company, I assumed full responsibility for translating this architectural pivot into a clear, usable interface layout. My day-to-day alignment occurred directly with the Product Management team to define product milestones, handle scope trade-offs, and transform feature requirements into user-facing systems.

I also synced periodically with the founder to present major strategic shifts and ensure design directions aligned perfectly with upcoming fundraising goals. Because the entire development team sat in a separate department, I built a structured design-to-development pipeline from scratch. I independently managed the handoff of engineering assets, defined functional interaction states, and ran systematic design QA directly inside their engineering sprints.

The parameters of this project were sharply defined by engineering limitations. Because the platform was live with active enterprise configurations, the product managers and development team had limited bandwidth for structural or architectural code refactoring. A complete database or workflow restructure was fundamentally off the table. Every deep UX problem had to be resolved purely through interface logic: manipulating visual hierarchy, engineering reusable component structures, perfecting typography scales, and building consistent layout systems without breaking existing URL paths or back-end API payloads.

To ground my strategy in regional competitive realities, I mapped the gaps across the current Indian AdTech landscape through three core market observations.

To ground the redesign in real buying behavior, I ran consolidated testing loops with participants drawn from across the Indian agency and brand landscape.

My consolidated testing loops exposed systemic multi-app fatigue. Users were explicitly overwhelmed by the requirement to manage assets across fractured product tabs, as captured in their direct feedback:

“The tools work fine individually, but constantly moving between them to download and upload files completely slows down my creative rhythm during busy client seasons.”

“Creating an ad asset in one menu and then setting up the budget inside another feels like restarting my entire workflow from scratch every time.”

When reviewing the live product that was running when I joined, I discovered three recurring UX problems that directly threatened user trust and system adoption.

The problem wasn't that the individual tools were weak. The problem was the massive operational gap between them. Users weren't dropping off because the generation engine failed; they were dropping off because downloading assets from a creative tool and re-uploading them into a separate campaign launcher created severe, repetitive context-switching exhaustion.

To fix this structural broken loop, I mapped out the consolidation of our core features into an interconnected flow.

Before committing, I weighed three strategic options against the creation-to-deployment gap and our engineering constraints.

This realization completely changed our approach. Instead of building alternative multi-step setup wizards, I recommended to the PMs and the founder that we replace the static homepage entirely. We shifted our focus to an architectural transformation: merging the text-to-image and text-to-video tools into a single multimodal application called Pixieo, integrating the autonomous single-prompt campaign infrastructure called Xeno, and tying everything together under a central chat orchestration layer called Neo.

To implement this without overloading our separate development team, I made a strict product design trade-off. I chose to postpone the creation of complex file management structures and asset sorting systems. Instead, I prioritized a clean, unified split-screen workspace that dynamically displays previews alongside live conversations, keeping developer implementation straightforward and efficient.

During our cross-functional alignment syncs, severe stakeholder tension emerged regarding layout execution.

I resolved this friction by introducing an expandable, persistent sidebar layout paired with responsive action chips. This strategy kept engineering implementation straightforward while completely hiding complex agent sub-processes behind clean, contextual confirmation states, satisfying user speed requirements while keeping technical validation available on demand.

When the product pivoted toward a conversational interface, my first concept design treated the chat input field as a floating modal window laid on top of the original grid interface.

Direct validation loops with the PMs and initial engineering check-ins quickly exposed the flaws of this approach. The floating chat layout felt disjointed and blocked users from seeing generation previews or underlying data tables. I rejected this overlay model and designed an integrated, split-screen architecture where the conversational panel acts as the main core interface, directly outputting interactive assets and data blocks right into the main flow.

Before committing to high-fidelity visuals, I mapped the conversational workspace and its supporting surfaces as low-fidelity wireframes, validating layout structure and hierarchy without breaking the existing engineering architecture.

This structural evolution outlines how we shifted from a fractured, multi-window layout into an integrated, single-context conversational hub.

A single user prompt enters the Neo Hub, which detects intent and orchestrates the rest of the system end to end.

To deliver a clean, customizable layout environment without introducing database-level structural updates, I engineered a flexible global navigation structure featuring a dedicated user pinning zone.

To demonstrate how the system responds to real-world tasks, here is the direct interactive path an agency director takes when setting up a live campaign.

Unlike traditional SaaS software, building a conversation-driven automated interface introduced unique behavioral issues that required intentional layout responses.

Problem: Deep conversational flows often filled the screen, causing users to lose track of their position within the wider application.

Insight: Users require a fixed visual baseline to navigate complex, multi-agent workspaces confidently.

Decision: I implemented a strict, immutable global sidebar layout that anchors core platform spaces, keeping global workspaces accessible at all times.

Problem: Forcing users to scroll down or swap windows to check generated assets disrupted their mental focus.

Insight: Conversational prompts and their visual outputs must exist within the same visual frame.

Decision: I structured a twin-panel layout model. The left area houses the conversational thread, while the wider right canvas displays live image previews and data grids instantly.

Problem: Performance marketers used separate optimization reports far more frequently than creative generation feeds.

Insight: Interface efficiency scales rapidly when operators can personalize their primary navigation environment.

Decision: I engineered a custom drag-and-pin area within the global sidebar, letting users pin frequently used features directly to their navigation tray.

Problem: Marketers routinely ignored automated suggestions because typical layout grids looked uniform and failed to highlight financial risk.

Insight: Automation insights require distinct visual grouping and explicit financial impact labeling to drive action.

Decision: I redesigned the recommendations view into a modular feed. Each card isolates its target optimization category, explicit data signals, and risk assessments, complete with a prominent command button.

To support long monitoring sessions and maximize visual clarity under variable agency lighting, I personally managed the platform accessibility overhaul to ensure full AA compliance.

I tested high-fidelity prototypes across three successive validation cycles with external agency media buyers and product managers. Testing exposed a fundamental architectural failure: when users ran deep conversational sequences within the central chat hub, the interface expanded to fill the entire screen. This completely hid the global navigation sidebar, causing users to lose track of their workspace context and feel trapped inside the chat loop.

To resolve this issue, I made the primary global sidebar persistent and introduced prominent agent state indicators that explicitly flash which system engine (Neo, Pixieo, or Xeno) is currently running. I also added a drag-and-pin mechanism that lets users select any deep tool module (such as an optimization feed or budget tracker) and lock it directly to their navigation sideboard, ensuring fast access and constant workspace control.

The final configuration achieves absolute visual clarity, balancing complex multi-agent automation with clean, scannable layout structures.

The operational value of this product architecture transformation was validated directly across internal databases, marketing logs, and partner tracking systems.

To provide believable validation to stakeholders without relying on arbitrary revenue assumptions, we measured the design overhaul success by tracking its direct impact on our enterprise sales pipeline velocity. By moving from a fragmented toolset to a unified conversational core, we drastically reduced prospect hesitation during live product demonstrations.

By framing the platform impact through clear conversion velocity percentages rather than speculative financial projections, the design system directly proved its worth as an enterprise sales acceleration engine. The premium feel of the interface allowed the product management team to advance high-value agency accounts through the evaluation pipeline far more predictably than the legacy layout allowed.

Early in the project, I assumed that performance marketers preferred dense, compact data presentation to maximize screen real estate. Based on this assumption, I designed an ultra-compact data table that packed all asset outputs and campaign metrics into a single screen view. Usability testing with our product managers and trial users quickly disproved this direction, as individuals found the dense layout overwhelming and struggled to identify key actions. I adapted immediately, introducing structured tab groups and clearer spacing that improved data readability without requiring backend engineering changes.

This project highlighted the immense value of designing within strict engineering boundaries. Being the only designer in the company taught me how to collaborate closely with product managers to validate user needs against business realities, while managing handoffs to a completely separate development department. I had to discard a few complex interaction concepts because they exceeded technical implementation capacity. This taught me a vital lesson in design restraint: a perfectly executed standard component is infinitely better than an overly ambitious custom interaction that fails in development. I learned to ground my creativity in technical reality, using precise spacing, typography, and contrast to solve complex layout problems.

The foundational design system established during this overhaul sets up a clear path for future product growth. As engineering bandwidth expands, the next development cycles will focus on introducing several capabilities.