The company
Next Core Revolution builds Next Core Suite, procurement software for complex institutions that have their own regulations and cannot change them to fit a tool. Each purchase runs under the institution's rulebook and ends with a complete, defensible file.
Large institutions buy under long, detailed rules, and most follow them with Word files, spreadsheets and email. Proving afterward that a purchase followed the rules is the expensive part. In our design the rulebook is the input: its thresholds, approval matrices and required documents become the system's configuration.
The execution layer, which runs each purchase deterministically and leaves everything auditable, is in production at a Chilean university, where it replaces Word, spreadsheets and email in the purchasing office. The AI that reads a rulebook and proposes that configuration is still under construction, and a person approves any configuration before it goes live. The company won Gearbox's 7th Gear Challenge in 2024, which came with a USD 25,000 investment.
My role
I am CTO and co-founder, and I have been building the product since 2023. I lead the technical side: the architecture, the engineering of the suite and its deployment inside the customer's own infrastructure.
The suite is built with Django and PostgreSQL on the back end and React with TypeScript on the front end, with single sign-on through OIDC and role-based access control.
Architecture
One product where the institution's rules are data, so each process is configured from its rulebook instead of being programmed by hand.
Access and security
Single sign-on through OIDC and roles that decide who can approve what.
Deployment
On-premise installation inside the customer's infrastructure.
What I have built there
Contracts portal for a university
Centralizes the document cycle of purchases and contracts: documents generated from templates, a multi-role approval flow, a versioned repository with full traceability, and semantic search. Single sign-on through the university's mail accounts and seven access roles. A retrieval-augmented AI with Gemini assists contract drafting and offer analysis, with embeddings stored locally in pgvector. It runs on the university's own servers, with no external cloud.
Diagnostic of a staff-contract process
Mapped a university's contract cycle for academic and administrative staff: over 400 manual reports a year, contracts handled by email and Excel. Designed the target process, with integration to the personnel system, electronic signature and registration with the labor authority, and a data architecture for sensitive information under Chile's data-protection law.
Operations platform for a retail chain
Sales control, inventory with automatic shrinkage calculation, demand projection and supplier price history for three branches, integrated both ways with the existing point-of-sale system over a REST API, plus a chatbot that answers operational questions in natural language.
Shared platform core
The common base of all projects: JWT and OIDC authentication, access control per endpoint and method, a multi-source integration engine, a versioned document repository, an AI module with retrieval-augmented generation and semantic search, and Docker, Nginx, Redis and CI/CD.
How it connects to my background
I have been on the buyer's side of this problem, in public hospitals where every purchase has to survive an audit. My thesis taught me the other half: take messy inputs, build explicit stages, and measure the output at every one. The suite applies that to institutional purchasing.