Build it once: why Nommo sits under every product
Sample post · for review before launch
Every Cr8r product needs search, memory and agents. So we stopped building them per product.
SearchProperties needs search. So does MeetLoop. So do ALOT and Wheels.ng. Each of them also needs memory, recommendations and, soon, agents that can act on a user's behalf.
The obvious path is to build those features inside each product. Follow it and you end up with four search stacks, four ways of storing context and four sets of bugs.
One layer, many products
Nommo is the alternative. It is a shared intelligence layer that sits under every Cr8r product. A product describes its data and its rules. Nommo handles the rest:
- Search that understands what someone meant, not just what they typed.
- Memory that carries context between sessions.
- Agents that can take an action, inside limits the product sets.
- Recommendations that learn from every product, not just one.
When we improve ranking for property search, event discovery gets the improvement too.
Why it matters for a venture company
A client project ends. Infrastructure compounds. Every hour spent on Nommo makes the next product cheaper to build and better at launch.
It also means Nommo can stand on its own. The same APIs that serve our products can serve other companies. That makes it a product, and possibly a company.
What's next
We are moving the remaining per-product search into Nommo first, then memory. We will write up what breaks.
Build first. Learn fast. Improve constantly. Scale what works.