FROM ZERO SELF-SERVE TO FIRST PQL IN TWO WEEKS — DESIGNING PLG FOR AN ENTERPRISE VOICE AI PLATFORM

Responsibilities

Product design, PLG analytics framework, user research (AE interviews, session recordings, behavioral analysis), PQL signal definition.

Client

PolyAI

Year

2026

Info

PolyAI builds voice AI agents for enterprise. The platformwas never public — every deal was sales-led. I designedthe company's first self-serve experience that deliversproduct value in under 3 minutes: a way for anyoneto sign up, build a personalized AI agent, and talk to it.

AGENT WIZARD

One input. One URL. The AI does the rest.

Step 1 — Paste your company website. The AI scrapes FAQs, help pages, and contact details to build a tailored agent. No forms, no configuration.

Step 2 — Choose a voice while the agent builds in the background. Dead time becomes a moment of personalization.

Step 2 — Choose a voice while the agent builds in the background. Dead time becomes a moment of personalization.

Step 2 — Choose a voice while the agent builds in the background. Dead time becomes a moment of personalization.

Step 3 — Talk to your agent. Suggested questions reduce blank-page anxiety. A secondary CTA leads to the platform for deeper exploration.

Step 3 — Talk to your agent. Suggested questions reduce blank-page anxiety. A secondary CTA leads to the platform for deeper exploration.

Beyond the Wizard

The wizard is the entry point. Converting signups into product-qualified leads required a full PLG system around it:

— In-product walkthrough to guide technical exploration without overwhelming non-technical evaluators — 14-day trial nurture sequence with behavior-triggered emails co-designed with marketing — Mid-trial feedback survey to capture qualitative signals alongside behavioral data — Pendo analytics dashboard tracking the full funnel from signup to PQL to demo request — PQL behavioral framework distinguishing breadth from depth engagement patterns

This infrastructure made the next section possible — identifying the real aha moment from real user behavior.

The wizard is the entry point. Converting signups into product-qualified leads required a full PLG system around it:

— In-product walkthrough to guide technical exploration without overwhelming non-technical evaluators — 14-day trial nurture sequence with behavior-triggered emails co-designed with marketing — Mid-trial feedback survey to capture qualitative signals alongside behavioral data — Pendo analytics dashboard tracking the full funnel from signup to PQL to demo request — PQL behavioral framework distinguishing breadth from depth engagement patterns

This infrastructure made the next section possible — identifying the real aha moment from real user behavior.

Defining the Aha
Moment

Defining the Aha Moment

Through funnel analysis, session recordings, and AE interviews, I identified the core engagement loop: edit → publish → call → hear the difference. 83% of users who published a change called immediately to test it. This cycle separated casual browsers from serious evaluators — and became the foundation for PQL scoring.

Two PQL profiles emerged from behavioral analysis: breadth users (10+ sections in one session — systematic evaluation) and depth users (multiple days, dozens of edits and calls — actively building). Our first PQL matched the breadth pattern: 15 minutes, 10+ sections, converted to SQL.

Through funnel analysis, session recordings, and AE interviews, I identified the core engagement loop: edit → publish → call → hear the difference. 83% of users who published a change called immediately to test it. This cycle separated casual browsers from serious evaluators — and became the foundation for PQL scoring.

Two PQL profiles emerged from behavioral analysis: breadth users (10+ sections in one session — systematic evaluation) and depth users (multiple days, dozens of edits and calls — actively building). Our first PQL matched the breadth pattern: 15 minutes, 10+ sections, converted to SQL.

ImPact

  • First PQL → SQL through the PLG flow within two weeks of launch

  • 9 demo requests from self-serve in the first two weeks

  • Wizard adopted across sales — demo prep dropped from 3–4 hours to under 30 minutes

  • PQL behavioral framework established, enabling proactive outreach to high-engagement users

Next projects

From pioneering a new category to $1M+ revenue — designing a cloud cost platform for finance and engineering teams

Designing the procurement workflow builder that helped Vertice close a $50M Series C