A defined job
The user begins with a role-specific question and a clear expectation for the resulting work product.
Selected work01 / 04
Level AI · 2025 - Present · Agentic enterprise applications
AI Workers
Led the product before, during, and after launch - from strategy and experience design through pricing, GTM, and adoption.
Product strategy across the experience, launch system, monetization, platform direction, and adoption loop.
Product surface
A role-specific worker combines a clear job, familiar enterprise context, and an inspectable output.
70+
active customer tenants in 30 days
3.5x
weekly active users in 10 months
75%
first-month user retention
25K+
platform runs
Product point of view
Enterprise AI earns adoption when it combines a flexible natural-language surface with governed tools, evidence, repeatable workflows, and a clear path from first use to habit.
Evidence · Product system
How one question becomes an operational deliverable.
The public architecture exposes the product problem underneath the interface: translate intent into coordinated research, governed tool use, and work a person can act on.
Reading the system
A governed system
Planning, research, tools, organizational context, and guardrails are coordinated behind the experience.
An inspectable output
The deliverable carries the context and evidence needed for a person to make the final judgment.
01 / Context
Read the system before shaping the product.
Problem brief01A
The problem worth solving.
CX teams had millions of conversations but still depended on dashboards, sampled reviews, and analyst queues to answer operational questions. A generic chatbot would not be enough: the product needed tenant-aware data access, reliable tool use, explainable outputs, enterprise permissions, and use cases that created repeat behavior.
02 / Decisions
The decisions that shaped the product.
A decision register connecting product direction to the evidence and constraints behind it.
D01Decision
Designed the product as a portfolio of focused workers plus a general-purpose analytical surface, giving new users a clear starting point without constraining advanced questions.
D02Decision
Led the end-to-end launch system: discovery, interaction design, roadmap, packaging, pricing and monetization, sales enablement, and customer adoption.
D03Decision
Built a measurement model for breadth, depth, retention, and account concentration so interventions targeted the customers with the highest leverage.
D04Decision
Authored an MCP-first direction that reframed 10+ siloed workers as four composable analytical primitives, exposed as tools with reusable skills and rules.
Tradeoffs held in view
The tension stayed visible. The choice made it actionable.
Guided entry points vs. an open-ended AI surface
Choice
Paired focused workers with a general analytical surface so new users had a clear starting point without constraining advanced questions.
Fast capability growth vs. a coherent platform
Choice
Defined an MCP-first direction that converged siloed workers into reusable tools, skills, and rules.
03 / System
Sanitized product and adoption system
A reconstruction of the product layers and measurement logic used to connect governed capability with repeat customer behavior.
Sanitized reconstruction
InputValue
01Layer 1
Enterprise CX data
02Layer 2
Governed analytical tools
03Layer 3
Skills, rules, and context
04Layer 4
Agent orchestration
05Output
Evidence-rich outputs
04 / Outcomes
What changed.
- 01Reached 70+ active customer tenants in a 30-day window.
- 02Grew weekly active users 3.5x over 10 months.
- 03Achieved 75% first-month retention among users who tried the product.
- 04Scaled to more than 25,000 platform runs across manual and automated workflows.
Next system / 02