Accelerating Time-to-Value by 40%:
Re-engineering the AI Workflow.
Redesigning the core automation engine to enable non-technical teams to build complex AI agents without writing code.
Impact Overview
Enterprise teams lost 60% of their time to broken, fragmented automation.
Simplifai needed a complete reimagination of its core workflow product — turning a developer-only tool into a platform any operations team could own.
Manual & Error-Prone
Teams spent hours on repetitive data entry, creating bottlenecks and errors.
Fragmented Tools
5+ disconnected systems with no unified workflow — every handoff was a risk.
No Self-Service
Every small change required engineering time, killing team velocity.
My Role
Product Design & UX Lead — driving research, IA, interaction design, and usability testing to shape product strategy.
Timeline & Scope
2 Years
2021 – 2023 · B2B SaaS · 5+ enterprise clients
Tools
The User
Sarah, an Operations Lead, was drowning in manual tasks. We mapped her frustrations to specific product capabilities.
Select a Pain Point
"Data entry errors causing significant rework."
Automated Data Capture
How We Got There
Step 01 — Research
100+ user interviews & contextual inquiries
Embedded within the Simplifai team for 3 weeks, interviewing operations leads across 8 enterprise clients. Uncovered the core tension: users needed power without complexity.
03 — Design Leadership
Beyond the Screen
As the only senior designer on the product team, I owned the design function end-to-end — not just the pixels.
Weekly design crits
Ran structured critique sessions with 3 product designers and 2 frontend engineers every Thursday. Introduced a feedback rubric (usability, visual hierarchy, accessibility, business alignment) that reduced revision cycles from 3+ rounds to an average of 1.4.
Mentored a junior designer
Paired with a junior UX designer joining the team in Q2 2022. Set a structured 60-day onboarding plan covering Simplifai's design system, design token discipline, and usability testing methodologies. They shipped their first solo feature within 10 weeks.
Stakeholder alignment process
Presented design decisions monthly to the VP of Product and CTO using structured decision logs — a one-page doc covering: problem, options considered, decision made, and success metric. Eliminated the most common stakeholder change requests by building alignment before high-fidelity.
Established the first design system
Initiated and built the Simplifai component library from scratch (no prior system existed). Defined a semantic token architecture, authored contribution guidelines, and ran a 2-day design system workshop with the engineering team to drive adoption. Led to 42% faster dev cycles.
03.5 — Product UI
From Research to High-Fidelity
Three screens that tell the full story — the dashboard that replaced an infinite canvas, the node builder that made AI automation intuitive, and the monitor that gave enterprise teams operational confidence.
My Flows
4 flows · 3 active
The linear step-builder that replaced an infinite canvas. V1's FigJam-style canvas paralyzed non-technical users — "where do I even start?". Switching to a structured list view with smart defaults drove a 40% reduction in onboarding drop-off in 3 weeks.
Turning 6 scattered tools into one mental model
IA research revealed users were context-switching between 6 disconnected tools to complete a single workflow. We collapsed everything into a single, hierarchical model.
Redesigned IA — Simplifai Dashboard
Result: Everything a user needs is accessible within 2 clicks from the dashboard. Zero context switching. 68% fewer navigation steps measured in usability testing.
The Flow Builder
A node-based canvas that democratizes complex automation. Built with a custom React engine — no engineering required.
Scenario: When a new lead enters the CRM, Simplifai instantly classifies intent, routes high-value prospects to sales, and drops others into a nurture sequence — all in under 2 seconds.
Hover over any node to see details
Design impact: This single canvas replaced the need for 3 manual approval processes. The visual programming model tested at a 92% task-completion rate vs 41% for the previous text-config interface.
rate (usability test)41%before
redesign
Results That Speak
42%
Faster Product Iteration Cycles
So What? Speeding up dev means significantly lower R&D burn.
100+
User Studies Conducted
So What? Validated decisions reduce post-launch rework.
20%
Improvement in CSAT Scores
So What? Happier users directly translated to contract renewals.
* 40% onboarding time reduction measured via pre/post Maze unmoderated testing (n=47 sessions, Q2–Q3 2022). 73% adoption rate tracked via Mixpanel event analytics comparing 90-day cohorts pre- and post-redesign launch.
Custom Design System
So What? Unified UI reduces technical debt & bugs.
export const theme = {
colors: {
primary: "#6366f1",
accent: "#8b5cf6",
},
spacing: {
sm: 4, // 0.25rem
md: 8, // 0.5rem
}
}
Reflections
What Failed & What I'd Change
What didn't work: the canvas-first architecture
V1 of the Flow Builder used an infinite canvas with free-form node placement — inspired by tools like FigJam. In usability testing, non-technical users found spatial freedom paralyzing: “Where do I even start?” We scrapped the canvas in sprint 4 and replaced it with a linear step-builder with smart defaults. Adoption in that cohort jumped 28% within 3 weeks of the switch.
If I started over: involve engineers earlier in IA decisions
The information architecture for conditional branching (if/else logic in workflows) was designed entirely by me before any engineering input. When we handed it off, the nested-condition model required a significant backend refactor. I'd run a technical feasibility session with the eng lead before finalising the IA for any logic-heavy feature. It would have saved 2 weeks of back-and-forth.