Milairo Labs · Independent Research

Milairo Labs · Independent Research

Researching the systems behind
AI-native software.

Milairo Labs is an independent research and experimentation project exploring AI, accounting architecture, persistent context, and software systems. It is not an advisory firm or a commercial services business.

Explore the Thesis → Projects & Research →
Independent
Research & experiments
Lab
Build · test · learn
Applied
Independent research
Miami
Headquarters

Our thesis

Software is moving from tools people operate
to systems that can operate work.

AI is beginning to do more than generate content or answer questions. It can interpret information, use software, coordinate workflows, and take action across systems.

This independent research project explores what happens when those capabilities are applied to real business operations, and what architecture, controls, data, permissions, and human judgment are required to make those systems actually work.

  • 01
    Understand
    Start with the domain, process, rules, data, exceptions, controls, and judgment underneath the work.
  • 02
    Build
    Turn that understanding into product behavior, deterministic logic, AI agents, integrations, permissions, controls, and human review.
  • 03
    Learn
    Deploy, test, audit real behavior, publish what we learn, and use those findings to improve the next system.

What we are exploring

Where AI meets real operations.

The research is organized around questions, not services. Accounting and finance are our deepest domain today, but the underlying systems problems extend across complex business operations.

01Domain ExpertiseRules · judgment · standards
→
02Product LogicWorkflows · permissions · exceptions
→
03Data + SystemsLedger · banking · integrations
→
04AI + AgentsReason · classify · orchestrate
→
05Human ReviewApprove · override · control
→
06OutputReliable · traceable · useful
AI-Native Finance

How accounting, banking, payments, reporting, and financial operations change when AI becomes part of the operating architecture.

Agentic Business Operations

How agents can execute multi-step work across systems while preserving permissions, controls, accountability, and escalation.

Process Intelligence

How workflows can be decomposed into rules, judgment, exceptions, and actions, then rebuilt around AI.

AI Product Architecture

How deterministic systems, models, data, memory, tools, and humans should work together inside real products.

Research process

Research through
building.

01
Study the Work

Understand the real process, domain rules, data, exceptions, failure points, controls, and judgment before deciding what technology should do.

Process mappingWorkflow designControls
02
Build the System

Translate domain understanding into product behavior, architecture, agents, integrations, deterministic logic, permissions, controls, and human review.

AgentsProduct logicControls
03
Test + Publish

Prototype, trace how the product actually behaves, test real scenarios, audit implementation against intended requirements, and turn what we learn into better products and useful research.

Codebase analysisAI-assisted devTesting
Researching reliable AI systems

AI systems need more than intelligence. They need controls, traceability, clear ownership, and a deliberate boundary between automation and human judgment.

AI suggests. The professional decides.
You sign the work — not the software.
Data governance matters.
Privacy, provenance and permissions are research priorities.
Audit-defensible by design.
Clear trail on every output — what changed, who approved it, and why.

Research & experiments

Ideas made real.

Independent prototypes, experiments, and writing are how the ideas are tested. The artifacts matter more than the pitch.

PACO dashboard
Independent Product Experiment
PACO

AI-native operating system for Puerto Rico small businesses, with accounting, banking, payments, tax and financial operations underneath one product experience.

View PACO →
PAOLA personal AI context system
Independent Research Concept
PAOLA

A personal AI context system for persistent memory across conversations, documents, models and tools. PAOLA owns durable context, current and superseded state, provenance, and a model-independent agent layer.

Memory · State · Provenance · Agents
RE/TERM market intelligence terminal
Independent Product Experiment
RE/TERM

Real estate market intelligence and analytics designed to make institutional-style data more accessible to investors and professionals.

View RE/TERM →
The AI-Ready Firm book cover
Research + Writing
The AI-Ready Firm

A practitioner's framework for moving accounting teams from AI experimentation toward deliberate systems, workflows and operating models.

View book →
The AI-Ready Firm book cover

Research + Writing

The AI-Ready
Firm

Published · March 2026 · Milairo Press

A practitioner's guide to AI in accounting. The three-tier framework for where most firms are, why they stall, and how to build deliberately. Written for any accounting professional who knows AI is coming but doesn't know where to start.

Author
Victor A. De la Cruz, CPA
Audience
CPA firms · Practitioners · Finance teams
What it covers
The three-tier AI framework · Workflow audits · Building deliberately
Available on
Amazon →
"The best financial products start with understanding the work, the judgment, and the exceptions underneath it."
Victor A. De la Cruz, CPA
Founder · Milairo Labs
Background
Big 4 · Public Accounting · CPA
Expertise
Business Systems · AI Product · Accounting & Finance · Process Design
What I build
AI Product Architecture · PACO · PAOLA · RE/TERM
Based in
Miami, Florida
LinkedIn
linkedin.com/in/vdelacruzcpa