I turn AI confusion into aligned execution.

VP of Product, twenty years shipping at scale, six years building production AI systems myself. I find the bet that moves the business — when the org is chasing the impressive one — and get everyone behind it.

The problem

Executive pressure to do AI, with no agreement on what.

Pilots that impress and never ship.

A team that can't tell which bets matter.

Selected work

Four engagements, in the order I'd want you to read them.

01
CPO
2016–2019
10MM MAU

OCC — Mexico's largest jobs marketplace

Twenty years of dominance eroding. The founder gone, the acquisition fresh, no product strategy, 45-day release cycles. Our in-house matching was precise and thin on coverage — and when the group stood up a central AI recommendations service, the data-science team resisted becoming its consumer.

I championed adopting it anyway, because coverage was what recruiters were paying for. Then I replaced the 45-day cave-build cycle with roughly fifteen autonomous squads, each on a deploy stack it controlled.

The harder work was the CTO. SEEK asked me repeatedly whether he could be worked with, and I gave the question real time before concluding he couldn't; my assessment drove the decision to remove him. I then ran product and tech on an interim basis and rebuilt trust from the inside — weekly 1-1s with every tech lead, moving them from waiting to be told what to build to owning their domains.

The culture was the deeper constraint: a rigid top-down hierarchy where openly critiquing a director was unthinkable, being merged across a border post-acquisition. My instrument was a weekly newsletter to both OCC and SEEK leadership that reported failed experiments as plainly as wins. It became the group's benchmark for product-team transparency — cited by senior executives and by the CEO as the standard he wanted other markets to match.

+30%
YoY applications
release velocity
1→20
product org
02
ADVISOR
ONGOING
TAX PRACTICE

A national financial-advisory firm

Executive pressure to adopt AI inside the Office of the CFO, and an entirely non-technical practice team being offered tooling nobody there could maintain.

I diagnosed that what they needed was sustainable no-code automation, built credibility with live demos instead of decks, and sequenced a three-phase roadmap: workbook roll-forward, return-to-provision reconciliation, trial-balance categorization.

100s
hours projected saved,
firm-wide

One AI-assisted review surfaced a material error in a live client workbook. The debate about whether this was real ended that afternoon.

03
ADVISOR
HEALTH, FITNESS
& WELLNESS

Three health & wellness organizations

The same pattern in three sizes: everyone wanted to talk about the AI product, and in each case the money was somewhere else.

NASM
$100MM ARR
CERTIFICATION

Advised the head of product at a fitness-certification firm on AI-assisted product strategy — which consumer AI opportunities to fund first, and how to frame the roadmap so the executive team would actually back it.

Herbal Medics Academy
4 WEB PROPERTIES

The founder was ready to fund a speculative AI build. I ran AI-driven analysis across four web properties instead, surfaced conversion and top-of-funnel gaps manual review had missed for years, and traded the build for a roadmap ordered by return. The unglamorous work paid better.

Healthgena
100+ STARTUPS
VISION 2030

Resident AI and product advisor for Saudi Arabia's Vision 2030 HealthTech accelerator — coaching founders on where AI belonged in the product, user need, value proposition, business model, and three-year feasibility.

04
SENIOR PM
2012–2016
25MM MAU

Funny or Die

A household-name comedy brand losing ground to BuzzFeed. Expanding into snack-size media and video recommendations drove a record 25MM MAUs — and then the metrics stopped meaning what everyone assumed.

Engagement was climbing while revenue stayed flat, because ad CPMs were collapsing industry-wide. A business monetized on ads couldn't win on time-on-site alone. I reframed the strategy from keeping users on our site longer to going where the audience already was, and built the mobile-apps team from zero to do it.

Among what shipped: Dips, a full-screen short-video app for comedy that anticipated the format TikTok would take mainstream, and the Weather app my leadership was lukewarm on until our lead Sequoia board investor singled it out as the standout concept.

25MM
record monthly actives
+15%
MoM time on site
+20%
YoY, from the
mixed-media shift

Why I still build

Building with AI changes the product development paradigm, and you can't set strategy against a paradigm you've only read about.

Prototyping is nearly free now, so the expensive parts moved: evaluation, context design, and the maintenance bill on anything the frontier will do better next year. Six years of shipping production systems myself is how I know which of those a given bet will run into — so I can tell a team where to invest, where to buy instead of build, and what to refuse.

Two of my own products, as the evidence.

Meldoria journaling interface: a daily writing prompt above two saved entries
A generated Meldoria chapter: an illustrated spire above the sea, an italic epigraph, and narrative text drawn from the user's week

LifemapHQ / Meldoria

A multi-step LLM pipeline: assessment webhook, psychometric scoring, multi-domain completions, dynamic PDF report generation. Then Meldoria — a week of journal entries returned as a chapter of your own story, written against your psychometric profile. 222 published articles from an automated content factory alongside it.

Big Invoice: a dark invoices dashboard showing $67,536.72 received of $71,372 invoiced, 95% paid, above a list of client invoices
Big Invoice: an estimates screen tracking draft, pending and accepted totals above a filterable list of client estimates
Big Invoice app icon: black wordmark on a yellow rounded square

Big Works / Big Invoice

iOS invoicing for the trades, built solo in Flutter. A million tradespeople sold bloated software that all looks like accounting — in the part of the economy least exposed to AI displacement. Big Apps. Tiny Price. Black and yellow against a category of blue.

Live on the App Store →
Diagram: nine separate chatbot scenarios of 38 modules each were retired and replaced by one shared core engine of 92 modules, fronted by seventeen thin webhooks of three modules each. Adding a persona went from cloning 38 modules to writing a 3-module webhook.

Adding a persona went from a 38-module clone to a 3-module webhook. Nine personas became seventeen while the system got simpler to maintain.

The lesson

The expensive part of an AI system is rarely building the first one — it's the maintenance bill on the tenth. Duplicating a working bot nine times is the fastest way to build and the most expensive way to own. Collapsing them into a shared core with thin per-persona routers is the decision that made the system survivable.

The estate

49 scenarios in production · ~22 services · one person
  • Assessment pipeline
  • Content factory
  • Agent layer
  • Memory tiers
  • Social video pipeline
  • Email & notifications
  • Payments

This is the part that changed how I think about org design. The question stopped being how many people a function needs, and became what the smallest team plus the right automation looks like.

Dispatcher Video Generator Publisher

Most recent build, shipped this month: a social video pipeline with separated dispatch, generation, and publishing stages.

Services

AI Opportunity Audit

I audit the org, the strategy and the products, and come back with where AI creates value here — and where it doesn't.

2–3 WEEKS
FIXED FEE

AI Transformation Roadmap

The sequenced plan, a working proof-of-concept, and the internal narrative that gets leadership aligned behind it.

4–6 WEEKS
FIXED FEE

Fractional Product Leadership

I run product for you — strategy, roadmap, the team's operating rhythm — until you're ready to hire the permanent seat.

1–2 DAYS/WEEK
FROM $10K/MO

Advisory Retainer

A standing call and async access, for a product leader who wants a second read on the AI decisions.

MONTHLY
FROM $4K/MO

About

Alan Seideman, outdoors on a bluff above the beach at sunset

I've spent twenty years learning to tell a real bet from an impressive one.

Sometimes that meant making the unpopular call and being right: adopting an AI service my own team wanted to build, and lifting applications 30%. Sometimes it meant killing my own work — I decided not to build a coaching chatbot because frontier models were going to outrun anything I could maintain.

I've also been on the other side of it. I shipped a follow-a-topic network as Twitter launched and the location check-in two years before Foursquare. Being early taught me what being right actually requires, and I've been selling that distinction ever since.

I work in Spanish as readily as in English — I ran a 400-person Mexican company's product organization in it — so I can lead or advise LATAM teams directly, with no layer of translation between the strategy and the people executing it.

Remote-first. MBA, University of Denver. B.S., Haverford.