Forward Deployed Engineer, Agent Commerce

Product.ai; Los Angeles (CA); On-site; USD 220,000 - 275,000; Full time

Benefits offered by this job

Relocation support available

Job summary

Product.ai is seeking a forward deployed engineer who goes to the customer, writes the integration end-to-end, prices keyed access, and closes the first paying accounts. You will work directly with AI shopping agents and developers, bringing real integrations into production and feeding learnings back into the product.

Based in Santa Monica, CA, this on-site role requires hands-on coding at the customer's desk, ownership of the revenue number, and close collaboration with Cortex rules, pricing,

Job description

You go to the customer, write the integration with them until it works, and make sure Product.ai gets paid when a machine does the shopping.

I'm Michael, founder and CEO of Product.ai. I want to explain what a forward deployed engineer means here, because it's a specific job, not a title we borrowed.

A forward deployed engineer goes to the customer and builds the integration with them until it works and pays. Here the customers are the teams building AI shopping agents and assistants, and the developers who buy verified commerce data through our keyed API and MCP, the protocol AI assistants use to call outside tools like ours. You write the integration code with the customer yourself. You price the access. You close the first paying accounts. You bring what you learn back into the product, every time, not once at the end of a quarter.

We run this company on Cortex, a written internal system of rules for what we sell and how. You open it every morning through Claude Code, and you read it before you sell, not once during onboarding and never again. That habit doesn't end after the first integration ships.

I've spent 17 years building commerce technology, bootstrapped, 100% founder-owned since 2009, profitable, no outside investors, no board. We're the verified truth layer for shopping. We work out what's actually true about a product and a price, for human shoppers and for the AI agents shopping for them. SimplyCodes is our first proof at scale: codes that actually work at checkout, roughly $22M in revenue, profitable, fewer than twenty operators outbuilding companies 10x our size.

Why This Role Exists

I opened this seat once already this year under a commercial title, and the framing was wrong. A commercial title asks someone to run a pipeline. This work asks someone to build, in code, at the customer's desk, until the thing they bought actually works.

AI agents are starting to do the shopping. When one needs to know whether a discount is real, it reaches for a data source. I intend that source to be us, and paid when it happens. That means writing real integrations with real engineering teams, not just proposing them.

I own the strategy. But I can't also sit at the customer's desk and write the integration while I build the company, so that part is yours. You'd be the first forward deployed engineer here, the person who turns \"agents should pay us\" into working code, signed accounts, and priced access.

This seat goes out to the customer, builds the integration, and owns the revenue number. Until a second commercial seat opens, the demos, onboarding and support for the developers and agent teams who come through that door are yours too.

The System You'll Need to Model

• Cortex, and the discipline that comes with it. Every day starts inside it. Before your first integration call, you read the internal rule documents that state what we will and won't sell.
• The rules aren't handed down from me. Any operator here can propose a change to how Cortex works, and since August those changes go live the same day, without waiting on my approval. The system is written by the people running it.
• A two-sided market that pays for certainty. Shoppers and the AI agents shopping for them consume verified codes, proof a discount is real before checkout; merchants pay to protect their brand. Price both sides without breaking either.
• The agent-commerce shift. The buyer is turning from a human into a machine. AI agents are starting to choose their data sources the way search engines once chose which pages to rank.
• The keyed API lane, now the only door. Anonymous API access ended in August, and the free legacy codes close behind it. The paid, keyed door is the one that's left, and you inherit it live, not a blank page.
• The affiliate engine, and the second line forming behind it. Affiliate networks fund the business today, concentrated among a handful of major networks. A second line is forming too: verified research, sold as a product.

If reading that energizes you, keep going. If it feels overwhelming or underspecified, this isn't the right fit.

What You Will Own

• The integration, written with the customer. You sit down with the engineers building an AI shopping agent or assistant, or the developer buying our keyed data, and you write the code that makes the connection work. No proposal substitutes for a working integration; you build it yourself, in the room or on the call, until it runs.
• Pricing and packaging for keyed access. You price the API and MCP access you just integrated, prototyping the pricing pages and account flows yourself before anything reaches an engineer. Fill the paid door with paying, keyed accounts now that the legacy free door is closing.
• The first paying accounts, closed by you. You own turning a working integration into a signed, paying account, then bringing what you learned about the customer's real needs back into how we package and price the next one.
• A co-signed charter and a first-quarter number. Within your first quarter, we co-sign a charter for this seat: one machine-checkable number, plus a written split of what you decide alone versus what you bring to me first. That number is paid, keyed accounts and the revenue they carry, from machines using us. The first quarter is measured on the accounts you close, because the dollars from a new integration arrive later than the signature.

You’ll use the craft you already own, integrations, pricing, closing, and grow into the layer above: pricing keyed access as the market matures, and writing the playbook the next forward deployed hire runs.

Who You Are

How you think. You think in systems and money at once, keeping a customer's stack, a pricing table, and the agent-commerce shift in your head while you reason from first principles when the map runs out. You update fast when the work proves you wrong.

How you work. You've personally wired an integration into a customer's stack, then closed the account yourself once it worked. You treat AI as a force multiplier: agents against your own pipeline and research, checked rather than trusted on faith. You move between writing code and closing a deal in the same week without dropping either thread.

What you've built. You've taken a technical product from zero to its first paying customers, and you can point to integrations you personally wrote and deals you personally closed. Maybe you were the first forward deployed or solutions engineer somewhere, building the connection and the motion from nothing, or you founded something, built it yourself, and sold it before you could afford an engineering team. API, developer, or platform revenue is familiar ground. We care about what you built and closed and how you reasoned about it, more than where you did it or how long your resume is.

Who this isn't for.

It's wrong if you need a quota, an accelerator, a comp plan, a defined territory: this is an ownership seat, and the number is yours to define and defend. It's wrong if you want to propose the integration and hand the build to someone else, or if you're counting on the system getting out of your way once ramped: the reading and the writing both come first here, and neither ever lets up.

It's wrong if the draw is a big-logo title, or if you only do one slice, just the code or just the close: you'll write the integration, price the access, and close the account, all at once. And it's wrong if your instinct is to manage upward and narrate progress. Here the working integration and the signed account are the story, and there is nowhere for activity to hide.

How We Evaluate

I don't run traditional interviews. Every stage here is demonstrated performance on real, work-relevant tasks.

• An async video screen. About 15 minutes, on your own time. I want to see how you reason, not how you present.
• Calls with company stakeholders. Short conversations with the operators closest to this seat.
• A conversation with me. How you model the integration and the deal together, where you push back, and whether you can argue the work with me, live, and not fold.
• A paid work trial. Real work, about two weeks, at a rate that respects your time: a real integration with a real prospect, the pricing model for keyed access, or the first outreach on a new account. I watch how you get grounded in Cortex, whether you can find the account, whether you assess your own work straight, and whether you're someone I want in the room.

If the work above reads like yours but your resume is unconventional, apply anyway. We hire on the work and the reasoning, not the pedigree.

Compensation & Ownership

Total first-year comp: $400,000 to $500,000. That's base, plus eligibility for the company's ownership and profit-share programs.

Base: $220,000 to $275,000, top of market for this seat.

Ownership: eligibility for the company's ownership and profit-share programs. Grants are performance-based, terms discussed at the offer stage.

Benefits: 100% family premium coverage. Effectively unlimited token budget, steered by return.

Based in Santa Monica, Los Angeles, in person, five days a week. Relocation support available for the right builder.