Back to NewsroomThe Business Case for Ontology-Driven AI
Mode.IQ, the mortgage ontology behind it, and why it compounds
Artificial intelligence has become very good at answering questions, summarizing documents and drafting content. Most lenders and servicers that have added it to their existing systems report incremental improvement. The results fall well short of what was expected.
The reason is that AI is only as effective as the context it understands. That context comes from an ontology. What follows explains what an ontology is, what Moder has built on top of one, and why that position gets stronger the longer it runs.
"AI is only as effective as the context it understands."
The friction nobody talks about
Ask a mortgage operations leader where their teams lose the most time. The answer is rarely underwriting decisions. It is some version of finding things.
A customer service representative moves between five systems to answer one borrower question. A processor hunts through policy documents to confirm a procedure. An underwriter emails the one senior colleague who remembers how an unusual scenario was handled three years ago. This is the quiet tax on mortgage operations, and it is a large part of why cycle times run long and compliance costs more than it should.
The friction is old. It became urgent when AI arrived, because AI inherits it. A model working from scattered, inconsistent or outdated internal knowledge amplifies compliance risk. It gives wrong answers confidently, at scale.
What an ontology is
An ontology is a structured representation of how a business actually works. It defines the customers, accounts, assets, documents, processes, events and policies that make up the company, and the relationships between them.
It settles the questions that every system in a lender’s environment answers differently today. What is a customer. What is a loan. How are they related. What has happened on this account. Which policy applies. Who made a decision, when, and why.
Underneath, an ontology holds six things: the objects a business reasons about, the links between them, the current state of each one, the logic and policy used to evaluate a decision, the actions a person or a program is permitted to take, and the permissions governing who may know, propose, approve, execute or override.
A database stores information. An application performs a function. A data warehouse holds a record of what already happened. An ontology creates a common understanding of the business across every system that supports it, and it updates as events occur. In most lenders today, the information exists and the understanding does not.
What that looks like in mortgage
Start with the loan. It connects to the borrower and to the property securing it. The property connects to a county, and the county connects to any hazard declarations in effect. The loan also connects to every call, payment, document, complaint and escrow event on the account.
Consider a borrower who falls behind on payments, calls with a question about a statement, asks for an escrow review, and files a complaint with a regulator. Today those are four events sitting in four systems. Connected, they are one situation with a visible pattern, available while there is still time to act on it.
What Mode.IQ is
Mode.IQ is the intelligence layer Moder operates above the systems a lender already runs. It is built on a mortgage-specific ontology spanning origination, servicing and compliance, developed on the Palantir Foundry platform under Moder’s exclusive strategic partnership with Palantir. Moder, an Archwell Capital company, holds the client contract and carries accountability for the result.
Governance is what makes this usable in a regulated business, and it comes down to three things a lender can check directly. Every recommendation is grounded in that lender’s current, approved policy. A person stays in control of the decisions that matter, with the system flagging, recommending and validating. Every action leaves a structured, evidence-backed record an examiner can review. Client data stays inside the client’s environment and is not used to train shared models.
A workflow that satisfies all three has governance built into it. A workflow that satisfies one or two has a gap, however capable the model behind it.
Why this reaches production
The conventional enterprise sequence is to unify the content, govern it, and then pursue value. That sequence is why many lenders are two years into a data and document remediation program with nothing running.
An ontology allows a different order. A curated knowledge repository is a snapshot, and it starts degrading the day it is finished. An ontology that the daily work writes back to stays current, because anything stale breaks a live workflow that afternoon. A lender does not need clean data everywhere to begin. It needs an authoritative source for the one decision being automated, and then the next one.
Value arrives one decision at a time, while the broader content and governance work continues in parallel.
What is running today
Moder has dozens of connected workflows in production across servicing operations, each built inside live operations with real users.
Escrow analysis is the clearest example. A full analysis cycle runs across tens of thousands of loans in a single pass, clears the accounts that need no further review, and routes the remainder through rule-based exception checks. Staff who once ran that process by hand now work only the exceptions the system flags. Tax exemption processing, call summarization with sentiment detection, complaint intake pulled directly from regulators and the call center, and ACH validation all run on the same underlying structure.
Why the position is durable
Mortgage outsourcing has been built on labor leverage for decades. AI puts real pressure on that model. Moder already runs mortgage operations at scale, which turns that pressure into an advantage, because the operating knowledge required to automate this work correctly sits inside the business.
Four things sit behind the position, and the combination is what matters.
- An exclusive strategic partnership with Palantir.
- Mortgage-specific logic developed through years of running origination and servicing operations, built workflow by workflow inside real client environments.
- A client base wide enough to carry the cost of that build and to apply each new workflow across many relationships.
- Accountability for the outcome. Moder carries the mortgage expertise, the workflow build, the governance and the result in a single layer. Outsourcing firms sell labor. Software vendors sell a tool and leave the integration work to the buyer.
A new entrant would need all four at the same time.
What compounds
Every workflow added, every exception an operator resolves and every client brought onto the platform adds to the same structure. Each new workflow is easier to design, govern and scale than the one before it, and the accumulated operating logic behind it cannot be bought quickly.
The structure also travels. Contact centers, document processing, underwriting and quality control work the same way whether the product is a mortgage, a car loan or a homeowner’s policy. Moder is extending into banking and insurance on the same platform.
For a lender, this lowers cost to serve and closes compliance gaps without replacing the systems already in place. For an investor, it is a business that becomes harder to copy the longer it operates. Both come from the same structure.