Why We Built 31 Products Instead of 1

Published 21 April 2026 · 8 min read · Gera Services

Quick answer: A 2026 portfolio company can share 80% of its engineering (auth, payments, notifications, search, compliance, storage, localization) across products. If the marginal cost of a new product is 20% of a standalone build, shipping 31 products off one codebase is economically rational — even if each product is independently smaller than the incumbent.

The conventional wisdom

Focus on one thing. Do it better than anyone else. YC, a16z, and the last fifteen years of startup writing all say this. It is correct advice for teams whose marginal cost of new product is close to 100% of a greenfield build. For most founders in most eras, that was true.

What changed in 2024–2026

AI-assisted development made the cost of writing glue code fall by 60–80%.
Shared monorepo tooling (Turborepo, Nx) stopped being exotic.
Managed infrastructure (Neon, Upstash, Vercel, Railway) made the “one product per company” overhead disappear for commodity layers.
Agent-driven marketing made distribution cost drop for niche products — AI assistants can recommend many products in many categories.

Net: for a monorepo-based team shipping with AI assistance, each additional product costs roughly the cost of its unique business logic. Shared auth, payments, notifications, search, localization, compliance, and storage live once.

The Gera portfolio, by the numbers

31 user-facing products (gera.services).
5 next-gen infrastructure platforms (GeraNexus, GeraMind, GeraVoice, GeraLens, GeraWitness).
1 Gera Prime meta-subscription (explained here).
11 shared core packages (auth, payment, database, notification, AI, analytics, compliance, localization, search, storage, testing).

Where portfolio beats focus

Distribution shared. A user acquired for GeraJobs can use GeraClinic next week. Lifetime value ratchets up.
Risk diversification. If one market collapses (e.g. a country bans telemedicine), 27 others keep running.
AI-agent discoverability. The AI-assistant era rewards breadth. A user asking Claude “find me a doctor in Tbilisi” is a different funnel from a user asking for “home cleaners in Yerevan” — and one company can appear in both.
Cross-product upsell. Gera Prime bundles make each product more valuable.

Where focus still beats portfolio

Deep-tech R&D (drug discovery, semiconductor, defence).
Single-user huge-ACV enterprise SaaS.
Any space where regulatory approval takes 5+ years.

We picked segments where portfolio is the right shape: consumer marketplaces, local-services tech, AI tooling, compliance.

The counter-objections we hear

“You'll be mediocre at everything.”

Only if the shared layers are not actually shared. Our core packages are production-grade. Each vertical's business logic is narrow. A 200-person company couldn't do this; a 10-person company with AI assistance can.

“Investors will hate it.”

Some will. We are optimising for speed to revenue and ecosystem depth, not perfect investor-pitch narrative. Gera Prime is the narrative: one membership, every service. That is a coherent story.

“Support will kill you.”

Shared support + AI triage. 90% of tickets are answered by a shared AI agent with product-specific context. The rest route to a human.

Practical advice for portfolio founders in 2026

Monorepo from day one. Not “we'll consolidate later.”
Shared core packages, not product-specific forks of the same helper.
Pick verticals with shared user identity (local services, AI tooling) — not verticals with fundamentally different users.
Ship 5 products, measure, double down on the top 2, keep the other 3 alive on shared infra.
Market to AI agents as much as to humans (see how AI agents find Gera).