AI agent onboarding design
Incorrect password, try again.
Qiuyi Yang home Say Hello

AI agent onboarding design

As the founding designer of a business startup, I turned the internal machine learning tool into an intuitive SaaS platform, focusing on agentic onboarding experience. Measure success by adoption, user activation, and user retention.

Impact
0→1
SaaS platform shipped
{{ actText }}
New users activate
{{ retText }}
New user retention
Duration9 months
My roleFounding Product Designer
TeamC-suite, PM, ML Engineers
PlatformWeb & Mobile SaaS

Salezilla is a lead-gen company that helps companies find hundreds of thousands of potential clients. It uses machine learning on the database instead of traditional filters, so results come back accurate and in larger volume.
When I joined, its clients were paying for a full service: Salezilla’s in-house specialists ran the tool and delivered results by hand. As demand grew, the company set out to turn it into a SaaS platform where startups get results by themselves.

Make an overwhelming tool feel welcoming

01

Before designing anything, I needed to see the problem the way a newcomer does.
The current internal tool is very technical.The first real result took about an hour of expert-level setup. 

Research
01
Experience audit
I walked the expert flow end to end and mapped every step a newcomer has to survive: prompt, 60+ labels, about an hour to a first result.
02
User interviews
14 sessions with external users and our own specialists, to see their needs and exactly where the two groups diverged.
03
Workshop
A 20-person cross-team workshop that gathered insights from stakeholders across every team.
04
Concept testing
I tested my first prototype, a traditional lead-gen flow, with external users. Specialists breezed through it, but most newcomers got stuck.
What I heard
“There’s so much on this screen. I don’t even know where to start.”
— External user, first session
“I skipped the instructions. I just wanted to click something and see what happens.”
— External user
“Oh, this is overwhelming. I’d need someone sitting next to me to get through this.”
— External user
The word I kept hearing:
“Overwhelming.”
CONCEPT TESTING INSIGHTS
01
Extensive instruction is the wrong way to teach
Users got through the flow easily once a step was explained in plain words.
02
People are willing to learn through an AI support chat
The same words they ignore in the UI, they happily read in chat, with far more patience.
Solution

A guided AI onboarding that talks in plain language

The agent greets you like a colleague, asks a few simple questions in plain language, and walks you through your first leads step by step. No manual, no setup maze.

01 Greets you like a human expert
The agent opens the conversation, explains what happens next, and sets a specialist’s tone from message one.
02 Labeling, kept light
A quick labeling round on a few companies teaches the agent what a good fit looks like.
03 Writes the prompt for you
Plain-language answers become an expert-quality prompt behind the scenes, so no one faces a blank box.
Baseline metric
0% {{ s1aText }}
completion rate
62 {{ s1bText }}
session average time for internal specialist
What this unlocked

Shipping it did one more thing: real users started flowing in, and with them quantitative tracking. For the first time we could see the drop-off stage by stage, and exactly where to aim next.
Session replay was blunt: only 17% of users made it through to a result. The goal or next stage is increasing the completion and activation.

Results from the start to earn user trust

02
Problem

The biggest leak was trust: a 34% drop in the first five minutes, before any result existed.

The flow let users spend 29 minutes until show any result, way past anyone’s patience.

Drop-off per minute against session time, peaking above 6% within the first five minutes
“I don’t trust your platform yet. I’ll look at the result, then decide if it was worth my time.”
- ONE USER, IN SESSION REPLAY
Solution
Back-and-forth chat, long waiting 10 questions · 29 min wait · 1000+ results
Results from the start 1 question · 5 min wait · 100+ results
Before — a long back-and-forth chat history before any results appear
6% {{ s2bText }}
flow completion rate
2% {{ s2cText }}
user activation
29 {{ s2aText }}
session average time to result

The AI tackles the hard task, user just approve

03
Problem

Prompt improvement is the key to accurate results, but it’s the hardest task, and most newcomers never did it. So 53% of them spent a long time and never got a good result.

Solution

Automate hard tasks, with human decision in the loop

The agent automates the experts’ move: when companies don’t fit, it refines the prompt with that learning. But rewrite quality isn’t perfect, so the user keeps the final call.


The design challenge was picking that moment: finding the right frequency and context where a manual confirm is necessary and feels effortless. 

01 A standing prompt hub

🟢 Low disruption
🔴 High cognitive load
First-timers never think to visit a settings-like page to fix accuracy.
02  In chat after each batch 
Shipped experience
🟢 Minimum disruption
🟡 Medium cognitive load
The right moment, the right amount of friction 
03 A nudge per rejection

🔴 High disruption
🟢 Low cognitive load
It interrupted the labeling rhythm users had just settled into.

Every batch of five labels feeds the next rewrite, so the prompt gets sharper the longer you label.

37% {{ s3bText }}
flow completion rate
26% {{ s3aText }}
user activation rate

Intuitive agentic experience across devices

04

As the founding designer I built the design system from scratch, across web and mobile. Focusing on copilot interaction that feels effortless and trustworthy.

01

A guided experience for the blank canvas dilemma

Users never face an empty screen. The agent opens with one clear ask: verify your company.
Every step arrives with its goal stated, and finished work slides away.

02

Intuitive inputs for all levels

For a difficult task like the prompt rewrite, there're 3 levels of effort, and every user finds the move that fits them.
- Regenerate: for a fresh version with no effort
- Open inputs: A non-tech user can quickly correct things in plain language
- Expert tweak: Clicks into the text and edits directly. 

03

AI transparency

Sometimes the AI can’t complete the task, so the design prepares for a graceful failure and the best next act. The agent keeps its thinking out loud to gain user trust.
The flow continues with whatever users give.