Miami-Dade is the proving ground. The larger ambition is a nationwide partnership model that combines local real estate expertise with AI-assisted decision intelligence built for much larger portfolios.
MIAMI — Real estate has never had more data. Yet some of the most important investment decisions are still made by experienced operators manually piecing together information from property records, ownership histories, mortgages, permits, transactions, rents, neighborhood trends and dozens of other signals.
The challenge is not finding information. It is understanding what matters, how the pieces relate to one another, and whether a seemingly ordinary property contains an opportunity that others have overlooked.
That is the problem Miami-based LIBA Capital is trying to solve.
LIBA is a real estate investment and asset management firm focused initially on Miami-Dade residential and small multifamily properties, including value-add, workforce housing and special situations. Its current portfolio comprises 20 units with an estimated value of approximately $3.6 million.
But the portfolio is also serving another purpose: it is the real-world laboratory for a much larger technology and investment thesis.
Behind its acquisitions, LIBA has been building a decision-intelligence system designed to help investors discover opportunities, connect fragmented evidence, underwrite properties, challenge assumptions and ultimately follow an investment from the first signal through operating performance.
Finding the story hidden inside the property
LIBA’s platform currently screens a universe of more than 12,000 Miami-Dade parcels. It brings together property characteristics, ownership, transaction history, financing, permits, rental information, neighborhood conditions, market indicators and risk signals that would otherwise need to be assembled from numerous sources.
The goal is not to create another national property-search portal or a chatbot that produces attractive investment summaries. LIBA is trying to build something closer to an investment operating system: a structured way to move from evidence to an investment thesis, from an investment thesis to underwriting, and from underwriting to a decision that can later be measured against reality.
“A property can look completely ordinary when you examine one record,” said Akos Balogh, co-founder of LIBA Capital. “Then you connect the owner, financing, transaction history, permits, rents and what is happening around it, and suddenly you may be looking at a completely different situation.”
The system is designed to surface those relationships and show the evidence behind them. But LIBA has deliberately kept the final investment decision with people.
“The model proposes. Principals decide.”
That rule reflects an important distinction in LIBA’s approach. Artificial intelligence can help examine records, identify relationships, compare evidence, expose inconsistencies and challenge an investment thesis. Core financial calculations — including NOI, cap rate, debt service, DSCR, cash-on-cash return and other underwriting metrics — remain governed by deterministic financial logic.
The objective is not to replace the investor. It is to give the investor a more complete, auditable and increasingly intelligent view of the decision.
The advantage of being your own first customer
LIBA has one advantage that many technology developers do not: it can test its conclusions by actually buying the property.
Before an acquisition, the platform can preserve the assumptions behind the decision — purchase price, renovation budget, timing, expected rent, financing, NOI and projected performance. After the acquisition, those assumptions can be compared with what actually occurred.
Did a renovation expected to take 90 days require 120? Did costs exceed the original underwriting? Did rents stabilize where expected? Was a risk correctly identified? Did the property outperform for a reason the original analysis failed to recognize?
“We do not want a system that becomes better at explaining why it was right,” Balogh said. “We want one that becomes better because it can see where we were wrong.”
That creates a feedback mechanism between data, analysis, principal judgment and actual investment outcomes. Over time, LIBA’s aim is for the system to build a decision memory: not only what the firm knows about a property today, but what it believed before a decision, what the principals chose to do and what happened afterward.
For an industry now flooded with AI products, that is a meaningful distinction. A model can generate a persuasive investment memo in seconds. It is considerably harder to remain persuasive six or twelve months after capital has been committed.
Miami-Dade is the laboratory, not the limit
LIBA deliberately began in a single market because real estate remains intensely local. The same headline metrics can mean very different things from one neighborhood, municipality or ownership situation to another.
Miami-Dade gives LIBA a controlled environment in which to go deep rather than broad: learning how local records behave, how individual submarkets differ, which signals actually matter and whether the intelligence improves decisions over time.
But LIBA’s ambition is not to remain a 20-unit Miami investment firm with an internal software tool.
The larger strategy is to develop a nationwide AI-enabled real estate partnership model capable of supporting substantially larger portfolios.
Instead of attempting to become the local expert in every U.S. city, LIBA envisions partnering with strong real estate operators who already possess the relationships, market judgment and execution capability that cannot simply be downloaded from a database.
LIBA would provide the intelligence infrastructure: data integration, opportunity discovery, underwriting, decision memory, portfolio intelligence and AI-assisted analysis. The local partner would provide what remains inherently local: experience, relationships, physical execution and final judgment.
“We do not believe the future is one centralized AI pretending it understands every block in America,” Balogh said. “The more interesting model is combining powerful intelligence with exceptional local operators — city by city.”
That approach could allow each market to develop its own local intelligence while operating on a common decision architecture. If it works, LIBA believes the model could scale from hundreds of properties to portfolios measured in thousands of units without losing the evidence, underwriting discipline and local context behind individual decisions.
From acquisition intelligence to portfolio intelligence
The opportunity also extends beyond deciding what to buy.
As a portfolio grows, the same system can potentially follow an asset throughout its life — due diligence, renovation, leasing, operating performance, refinancing and disposition. That changes the question from simply “What should we acquire?” to “What should we do next with every asset we already own?”
Which properties are outperforming their original underwriting? Which assumptions are deteriorating? Where is renovation capital producing the best return? Which rents are below achievable levels? Which asset should be refinanced, improved, held or sold? Which earlier prediction proved wrong, and should that change the way the next acquisition is evaluated?
Those questions become exponentially harder as a portfolio grows. LIBA’s long-term thesis is that decision intelligence can give principals the ability to retain institutional discipline even as the number of properties, markets and data points expands.
The underlying LIBA real estate intelligence methodology is the subject of a U.S. patent-pending application. Certain decision-intelligence technologies used by LIBA are licensed from Colle, a technology company also co-founded by Balogh.
A bigger bet than an AI property search
LIBA is still early, and the company is treating Miami-Dade as a proving ground rather than declaring the technology finished. But the ambition is considerably larger than building another software feature for real estate investors.
The bet is that a real estate organization can eventually develop a memory across thousands of decisions: what it knew, what it predicted, what its principals decided, what actually happened and what the organization should learn from the difference.
If that can be combined with strong local partners across multiple cities, the result would look less like a conventional property database and more like a distributed real estate intelligence network — one capable of helping investors find, evaluate and manage opportunities at a scale that would be difficult to reproduce manually.
There is also a useful discipline built into LIBA’s experiment. Real estate eventually forces every prediction out of the computer and into the physical world.
A model can rank 12,000 parcels. It can process enormous amounts of information and produce an elegant recommendation. But eventually someone has to sign the contract, wire the capital, renovate the property, lease it and operate it.
“There is nowhere for the technology to hide,” Balogh said. “Eventually we know what we paid, what we spent, what the property produced and whether we made the right decision. That is what makes this so interesting to build.”
Miami-Dade is where LIBA is teaching the system. The larger question is how far that intelligence can travel.
About LIBA Capital
LIBA Capital is a Miami-Dade-focused real estate investment and asset management firm combining principal-led investing with proprietary real estate intelligence. The firm focuses on residential and small multifamily opportunities and is developing a partnership model intended to bring its decision-intelligence infrastructure to additional U.S. markets.
Website: www.libacapital.com
