AI Product Builder · Growth and PLG
Building AI products that grow themselves.
I build AI products and the agent systems companies run on, then run the growth that turns them into revenue: 12+ years of PLG, activation and monetization. Fintech to cybersecurity, early-stage to 10M+ daily active users.
Learn more
12+ years in product growth, most recently building AI and agentic systems in production across AI SaaS, fintech, cybersecurity, and AdTech. Based in Tel Aviv. Most recently an in-product AI analyst on an agentic architecture, and before that an LLM assistant that became one of its company’s top sellers. I build AI products that people actually adopt, then own the growth number that proves it.
My focus is the space between product and revenue, end to end: acquisition, activation, retention, and monetization, plus the analytics stack underneath. Lately that means agentic products: systems that act for the user, not just answer them. My work spans B2B SaaS and B2C at consumer scale, from Chargeflow and Workiz to ReasonLabs.
From seed-stage startups to enterprise platforms at scale. I chase what moves metrics, not what fills slide decks.
Agents changed what a product is and what a team is. I build both, and own the number each one exists to move.
AI & Agentic Products
Taking AI products from concept to production. I define the use-case portfolio, design agentic architectures on MCP, and test every scenario against real account data before it ships, so the model's answers stay grounded in what is actually true for each user. The evaluation suite is part of the product, not an afterthought. I build hands-on with LLMs and multi-agent systems.
Agent Systems & Automation
Putting agents to work inside the company, not only in the product. I design the automation layer: subagent teams with their own tools and permissions, skills that encode how the work is actually done, MCP connections to the systems of record, and the human review points that keep it accountable. Built for SMBs, startups and enterprise, where the constraint is rarely the model and almost always the workflow around it.
Activation & Onboarding
Diagnosing and fixing the gap between signup and first value. AI products face unique activation challenges: trust in automation, longer time-to-value, and onboarding flows that ask for access before they have earned it. I map the entire journey from install to activation, identify where users drop off and why, then design and ship the flows that move them forward, whether the evaluator is a person or an agent.
Monetization for AI Products
Pricing that survives contact with inference costs. Seats stop making sense when an agent does the work, free tiers meet a bill that scales with usage, and outcome-based pricing needs an eval suite before it can be sold. I have owned monetization, billing and payments end to end across Stripe, Shopify Payments, and WooCommerce, and I treat pricing as a product surface rather than plumbing.
Evals & Trust
An agent that acts on a customer's behalf cannot be judged by eyeballing it. I build the grading layer: the eval suites that decide whether a release ships, the handoff checks that catch what step-level tests miss, and the guardrails that turn autonomy into something a business can sign off on. Trust is the adoption constraint for AI products, and it is measurable.
Growth Architecture
The growth architecture that turns adoption into revenue: product analytics, lifecycle, attribution and the warehouse they run on, connected into one loop instead of 5 disconnected tools. Metrics definition through execution, so the number that proves the work is instrumented before the work ships.
What I've built and shipped across AI-native B2B SaaS, fintech, cybersecurity, and AdTech.
Every step leaks a little. Lift one and everything after it grows.
Start 100,000 visitors. At rates a real self-serve funnel runs, 400 become paying customers.
Most AI products work in the demo and stall in production. I write about where they stall, and what closes the gap. 31 essays so far.
AI-Led Growth: The Motion After Product-Led
Half the industry says PLG is dead because the next user is an agent. Both sides argue about the wrong layer. The economics survive, the funnel's assumptions die, and the build list is the same either way.
The Company Operating System
A persistent agent that holds what the company knows and gets asked before a person does. Context coverage as a tracked metric, why summarizing transcripts costs you recall, and the agent as gatekeeper.
The Cheapest Token Never Arrives
Spotify routed bulk file reads to a cheaper model and reported around 90% token savings. The number is workload-shaped. The finding underneath it, that soft rules do not hold on agents, is not.
Agent Teams for a Product Org
An agent team is a permission model with names attached. Which subagents a product org actually needs, what each one is allowed to touch, and the handoff where quality quietly dies.
Automate the Handoff, Not the Task
Most AI use-case lists are task lists, and tasks are not where the time goes. What to automate for support, sales, finance, R&D, product and leadership, and why access is the real constraint.
Skills Are Institutional Memory
A prompt is typed once. A skill is how your company does the work, written down so an agent can execute it. What belongs in one, what rots, and why the first draft is a process audit.
Context Engineering for PMs
What enters the context window decides what your product can be right about. The 4 strategies, the failure mode of each, and why retrieval is confident about absence.
Buy the Agent or Build the Grader
Build versus buy is the wrong frame. The question is who owns the pass bar, why the framework choice matters less than anyone thinks, and the wrap option nobody puts on a slide.
What an AI Automation Is Actually Worth
The same build, 3 prices, and the reason is not leverage. Time removed from a queue, what that time costs, how often it runs, and the review burden nobody quotes.
When Your Next User Is an Agent
A person who hits friction retries tomorrow. An agent fails silently and takes the next tool in the registry. 3 kinds of users now move through the funnel, and each needs its own definition of activated.
Production Is an Org Problem
Almost every company has an agent pilot. Far fewer have one in production. The surveys disagree on the spread and agree on the causes, and model quality is not among them.
The Agentic Growth Team
Agents already run the middle of the experiment loop. What breaks when they do, what humans keep, and why the metric definition is now the highest-leverage artifact on the team.
Evals Are the New QA
Every AI product works in the demo. Production is a stranger asking about their own data. The scenario library is the new PRD, production data is the only honest test set, and the eval suite belongs in the release path.
Who Grades the Grader
Once a suite outgrows hand review, a model grades the model. That judge fails the same way the product does, confidently and fluently, and almost nobody measures it against a human first.
You Can Only Loop What You Can Grade
Anthropic's Claude Code team sorts loops into 4 types. Read the table and it is not a capability ladder, it is a delegation ladder: the check, the stop condition, the trigger, the prompt.
When Resolved Becomes a Price
Bill per resolved outcome and the rubric deciding what counts as resolved stops being a quality artifact. It becomes the billing engine, and an eval failure becomes an invoice that never goes out.
Pricing After Seats: What Growth Teams Own Now
The human seat was the unit of SaaS pricing for 2 decades. Agents broke it. When the unit of work detaches from the unit of pricing, the pricing page becomes a product surface.
The Free Tier Meets the Inference Bill
Free was a marketing expense that rounded to zero, until every action acquired a model call behind it. Free moved from duration to quantity, and now it needs a budget and a conversion job.
The AI Trust Gap: Why Traditional Onboarding Fails AI Products
AI automation asks users to trust a black box with real business outcomes. That trust gap creates activation friction traditional SaaS onboarding patterns cannot solve. What works instead.
Friction Was Doing a Job
In-assistant buying removed every step between intent and purchase, and converted worse than sending the shopper to the merchant. The steps that got deleted were carrying the decision.
The Transparency Tax
Disclosing chatbot identity before a sales conversation cut purchase rates by 79.7% in a field experiment. The timing fix that recovered it is what Article 50 of the EU AI Act removed on 2 August 2026.
The Moat That Exported Itself
Every deck calls memory the last AI moat. Then the 3 companies with the most of it to protect all shipped memory portability inside a month. Whatever survives a plain-language export was never a switching cost.
Attribution After the Click
Discovery moved inside LLM answers, where the click never fires. Share of answer replaces share of shelf, branded search becomes the shadow of invisible recommendations, and some influence is now honestly unmeasurable.
The Compound Effect of Small Funnel Fixes
Growth teams chase 10% wins and miss the ones that compound. A 5% lift in 3 places does more for ARR than a single redesign, and it ships in a week.
Distribution Is a Product Skill Now
The best product no longer wins by default. Building got cheap, discovery compressed into AI answers, and owned audience became an asset a product team has to build on purpose.
Metrics That Move Teams, Not Just Dashboards
Most product dashboards measure what is easy to count. The metrics that change behavior are the ones a PM, designer, and R&D lead can argue about in the same room.
Growth Runs on the Warehouse Now
Every tool in the old growth stack kept its own copy of the customer, with its own definitions. The copies are dying, and the team that runs on the warehouse owns definitions instead of tools.
Building Retention Before You Need It
Churn is a receipt, not an event. Retention is an instrumentation problem long before it is an analysis problem, and the cancellation flow is the highest-signal surface most products waste.
The Buyer, the Admin, and the User
The flow is designed for 1 person and 3 different people show up, each with a different definition of value. Detect the persona at entry, branch the first session, treat activation as a chain.
Going Enterprise Without Killing Self-Serve
Moving upmarket is usually framed as outgrowing self-serve. Backwards. Self-serve is the qualification engine the enterprise motion runs on, and the handoff decides whether you capture the deal.
Time to First Dollar
A marketplace has 2 activation problems running on different clocks. Supply activates on first earning, demand on first satisfying transaction, and referral loops built before either one works amplify churn.
Let's connect.
If you're building an AI product, or growing one, I'd like to hear what you're working on. Growth and PLG is where 12+ years of my work sits, so it is usually the fastest place I can be useful. I also take on a small number of advisory engagements.