Home › How to run 10+ products with a small team
Playbook · Portfolio operations
How to run 10+ products with a 5-person team (2026 playbook)
The multi-product problem.
At three products, a founder can hold the whole company in their head. At five products, they can't. At ten, the wheels come off.
The failure mode is predictable. The team keeps hiring to keep up. Each new hire adds coordination cost. Meetings multiply. Status reports multiply. Slack channels multiply. By the time the ops team is 20 people, the founders spend most of their time in sync with each other rather than shipping.
This is why most portfolio operators cap out at 3 to 5 products. The 5th product breaks the coordination model, and the team either shrinks the portfolio or grows into a mid-size company they didn't want to become.
There is a third path. It requires giving up two assumptions: that humans need to be the memory and that humans need to be the routing layer.
The pattern that actually works.
Studios that successfully run 10 to 20 products with 2 to 10 people have three things in common:
- One shared brain. Not per-product Notion spaces. Not per-team Slack channels. One place where every decision, customer conversation, ship event, and signal lives. Every teammate and every AI agent reads and writes to the same place.
- Cadences instead of meetings. A morning brief (auto-generated). A weekly signal review (30 min, async pre-read). A monthly portfolio review (2 hours, live). That's it. No standups per product. No product-team-syncs. No cross-functional-alignment-check-ins.
- An AI teammate that handles the floor. Draft PRs from error reports. Draft PRDs from customer feedback. Answer support with full portfolio context. Publish content after critique. Send follow-ups. Track anomalies overnight. The humans approve high-leverage moves; the rest runs.
If any one of these three is missing, the model collapses back into "we need to hire more people."
The systems in detail.
The shared brain
Every operator running multiple products hits the same wall: knowledge lives in heads. Product A's context is in the founder's head. Product B's customer feedback is in Sarah's inbox. Product C's Sentry errors are in a Slack channel nobody reads.
The shared brain solves this by making memory a first-class system. Slack + Drive + Gmail + Teams + WhatsApp + error logs + support tickets all flow into one structured memory. Tagged. Time-stamped. Source-attributed.
Two things matter: the memory has to be complete (partial memory produces worse decisions than no memory) and the memory has to be queryable (a memory nobody reads is a graveyard).
Products that give you this: Qualia, Sentra, Hyperspell. Only Qualia acts on the memory as well. Sentra and Hyperspell surface it to other tools.
The cadences
The three cadences replace roughly 20 hours of meetings per week for a 5-person team:
- Morning brief (10 min read, auto-generated). Every metric across every product. Anomalies flagged. Cross-product patterns surfaced. What shipped overnight. Open blockers. The founders read this over coffee.
- Weekly signal review (30 min, async pre-read). What did external signals (Reddit, X, competitor sites, App Store reviews) tell us this week? Which of them turned into evidence for existing product decisions? Which of them should trigger new work?
- Monthly portfolio review (2 hours, live). Which products are compounding? Which are flat? Which should scale, which should kill? What's the next product to ship?
That's it. No daily standups. No product-team meetings. No cross-functional syncs.
The AI teammate
The floor is the operational work that happens whether the humans are there or not: support tickets, error triage, content drafts, market research, PRD authoring, metric tracking, ASO metadata, paywall iteration.
At portfolio scale, this floor is enormous. Support tickets alone for 15 apps easily reach 50 to 100 per day. Error reports from Sentry across 15 apps: dozens per day. Content that needs to ship: at least one piece per product per week.
If humans handle all this, you need 20 to 30 people just for the floor. If the AI teammate handles it, you need 0 additional humans; you need one AI employee that spans every product.
The category is called AI employee or AI COO. Qualia is built specifically for portfolio operators.
The Sentez example.
Sentez runs 15 mobile and web products with 8 people. The AI teammate (Qualia) handles:
- 121 external signal sources scanned every day
- Morning executive brief written every day
- Draft pull requests from every Sentry error
- Customer support responses grounded in portfolio context
- Weekly content schedule across every product
- ASO metadata iteration
- Paywall variant analysis
- PRD drafting from customer feedback
The founders no longer write status reports. They read the morning brief, approve the high-leverage decisions, and ship.
How to actually start.
| When | What you install |
|---|---|
| Week 1 | Consolidate memory. Pick one shared brain. Connect Slack, Drive, Gmail, error logs, ticket queues, meeting recorders. Give it 2 weeks to learn your company. |
| Week 3 | Start the morning brief. Every day, the shared brain writes a 10-minute-read summary. Iterate on the format for 2 weeks until it becomes your first coffee. |
| Week 5 | Cut meetings. Kill daily standups. Cancel cross-team syncs. Replace with async Slack threads with the brain in-loop. |
| Week 7 | Turn on the action layer. Let the AI teammate draft PRs, respond to tickets, publish content in review mode. Approve everything for 2 weeks to calibrate quality. |
| Week 9 | Auto-approve routine actions. Let the floor run without you. |
| Week 12 | Add your 6th product. Then your 7th. The system scales sub-linearly with your team, not linearly. |
Total time to shift the operating model: about 90 days.
Common mistakes.
- Trying to keep old rituals. If you keep the daily standup after installing the shared brain, the shared brain gets stale.
- Buying memory-only tools. Sentra and Hyperspell are excellent memory. But if you don't have engineering capacity, you're paying for infrastructure you can't leverage.
- Adding people instead of systems. The instinct to "hire an ops person" is often the wrong instinct.
- Splitting the brain per product. One shared brain, not one per product. Cross-product signal correlation is where portfolio scale actually pays off.
Frequently asked questions.
How small can a team be and still run 15 products?
Practically, 3 to 10 people. Below 3, one founder becomes a single point of failure. Above 10, coordination cost starts to rise.
Can I use Notion or a wiki as the shared brain?
Not really. Wikis are for humans reading. A shared brain has to be readable by AI agents, structured, tagged, and connected to your comms tools.
Do I need engineering capacity to make this work?
No, if you buy an AI employee that ships pre-configured. Yes, if you buy just a memory layer.
How much does this cost?
For a 2 to 10 person team running 3 to 20 products: a portfolio-priced AI employee runs a few hundred to a few thousand dollars per month. Compared to the 20 to 30 headcount you'd otherwise hire, the math is not close.
Which product should I try first?
For portfolio operators, Qualia is built for this topology. For enterprise memory infrastructure, look at Sentra. For AI-augmented engineering only, look at Hyper.
How long before I see results?
The morning brief shows value on day 1. Cadence and floor shifts take 90 days to fully install.