Skip to content
Sei
All Posts
Ramkumar Venkataraman

Ramkumar Venkataraman

CTO & Co-Founder

92 articles

AI Agents

Scopes, Approvals and Records: Designing Consent for Personal Agents in Mortgage and Banking

A technical walk through the consent layer of the Personal Agent Protocol draft 0.3 — delegated sign-in, scopes and scope hints, operations with terms and expiries, approval records and token lifetimes — and how a mortgage lender or a bank would map its own consequential actions onto it: credit pulls, rate locks, application submission, transfers, payment plans and disputes.

6 min read
Read more
AI Agents

How You Test an Agent That Runs for 30 Days Before It Touches a Live Loan: Golden Files, Backtesting, and the Eval Harness SR 11-7 Actually Wants

A model that scores a loan in 300 milliseconds is easy to validate. An agent that holds a goal across weeks, re-plans as the file changes, and makes dozens of decisions along the way is not, and the usual accuracy-on-a-test-set approach misses most of what can go wrong. A technical account of the eval harness we run, the golden loan files we maintain, the backtest against historical underwriter decisions, and how ongoing monitoring under SR 11-7 and the NIST AI RMF applies to something that behaves more like a colleague than a classifier.

7 min read
Read more
Mortgage

Taking the Appraisal Waiver, or Not: The Value-Acceptance Decision an AI Agent Should Make and Document

When the automated underwriting system offers value acceptance, the lender gets to skip the appraisal, and gets to own the collateral value with full representation-and-warranty liability for the life of the loan. That is a decision, not a default, and most shops make it by reflex because the waiver is faster and cheaper. How we build an agent that decides whether to accept the offer on each file, documents why, and knows the difference between value acceptance and value acceptance plus property data, without ever touching the number the way an appraiser would.

5 min read
Read more
Mortgage

The Forged Document Problem AI Made Worse, and the Detection Desk That Answers It

A paystub edited in a browser tab used to need a steady hand. Now it needs a prompt. The volume of altered and fully synthetic income and asset documents hitting mortgage intake is rising because the tools to make them got free and good, and the human eye that used to catch the clumsy ones is no match for the clean ones. What a document-authenticity agent actually checks beyond reading the numbers, why the detection model is itself governed under SR 11-7, and where catching a forgery becomes a SAR obligation rather than a decline.

6 min read
Read more
Compliance

Governing Personal-Agent Traffic in Mortgage: Model Risk, Fair Lending, UDAAP and the Audit Trail

When borrowers' agents start talking to a lender's systems, the lender has a new channel to govern — one where the other party is software, the lender's own agent may be answering, and the record of what happened is easy to lose. How existing frameworks apply: GSE AI governance requirements, model risk and third-party risk guidance, ECOA and fair lending, UDAAP, and record retention under Regulation B and Regulation Z.

4 min read
Read more
Mortgage

Underwriting DSCR Investor Loans with AI: The Business-Purpose Determination That Keeps the Loan Out of Ability-to-Repay

A DSCR loan lives or dies on one classification the underwriter makes before touching the ratio: is this credit primarily for a business purpose, and therefore outside Regulation Z's ability-to-repay rule, or not. Get it wrong and a loan you priced as exempt is now a consumer mortgage with no ATR determination behind it. How we build an agent that computes the coverage ratio and, more important, documents the business-purpose call under 12 CFR 1026.3(a) so an investor and an examiner can both follow it.

6 min read
Read more
Mortgage

Dual Contracts, Hidden Seller Credits and Cash Back at Closing: Mortgage Fraud in the Transaction Itself

Some mortgage fraud is not about the borrower at all. It is about the deal: a purchase price that changed after signing, a seller rebate paid outside closing, cash back disguised as a repair credit, a second lien nobody disclosed, a seller who does not own the property. How each scheme shows up across the contract, the appraisal, the Closing Disclosure and the bank statements, the interested party contribution limits it is designed to evade, and the tie-out an AI agent runs on every file.

6 min read
Read more
Mortgage

When the Applicant Is an AI Agent: What Changes at the Mortgage Point of Sale

Personal agents are being built to find a home, compare lenders and fill in a mortgage application for the borrower. The regulations underneath that application did not change. A practical account of where an agent-submitted application starts the TRID clock, which steps still need the borrower's own consent, and how to design intake so an agent can do the work without making the borrower's decisions.

5 min read
Read more
Servicing

The P&I You Remit Is Not the P&I You Collected: Custodial Reconciliation With an AI Servicing Agent

Investor reporting and custodial accounting is the servicing function that never calls the borrower and never makes the pitch deck, and it is where a servicer loses its approval. How an AI agent runs the daily reconciliation Fannie Mae's Servicing Guide requires across P&I and T&I custodial accounts, why the scheduled/scheduled rule forces more accounts than people expect, and the line between reconciling the money and moving it.

6 min read
Read more
Mortgage

Appraisal Fraud and Illegal Property Flipping: Detecting Inflated Values Before the Loan Funds

Collateral fraud is the most expensive kind a lender can fund, because the loss lands on a property that was never worth the loan. How value inflation, cherry-picked comparables, condition misrepresentation, edited photos and flip chains show up in a mortgage file, what FHA's anti-flipping rule at 24 CFR 203.37a and appraiser independence under Regulation Z require, and the collateral checks an AI agent can run on every appraisal rather than a sample.

6 min read
Read more
Mortgage

Rental Income Is Three Different Numbers and Only One of Them Qualifies the Borrower: AI on the Investment-Property Calculation

A lease says one number, the appraiser's Form 1007 says another, and Schedule E says a third, and an underwriter has to pick the right one under rules that change by occupancy and by how long the borrower has owned the property. How an AI agent produces an auditable rental-income figure under Fannie Mae B3-3.1-08, where the 75 percent factor applies, and the reasonableness call the underwriter still owns.

6 min read
Read more
Servicing

Mortgage Fraud After Closing: Loss Mitigation Fraud, Equity Skimming, Short-Sale Flopping and Payoff Diversion

Fraud does not stop when the loan funds. Servicers see fabricated hardship packages, foreclosure rescue schemes that strip a distressed homeowner's equity, short sales to related buyers who resell days later, landlord insurance on loans closed as owner-occupied, and payoff funds redirected by a forged request. How each scheme shows up in the documents a servicer already receives, the rules around them, from Regulation X to the MARS rule, and where an AI agent fits without slowing down a borrower who needs help.

6 min read
Read more

You Ain't Seen Nothin' Yet

Book a Demo

Pack up some of your complex historical files — any loan type, any investor. We run them through intake, income and condition clearing, and in 30 minutes you see every condition we created and cleared efficiently for your own team, and why.

  • Any loan type, any agency guideline or custom investor overlays.
  • Every finding cited to the guideline or document it came from

Case study highlight

Loan review fell from 4 hours 50 minutes to 47 minutes per file, and 78% of items now clear at high confidence — so QC touches only the exceptions.

Pre-Close QC Agent · Better Mortgage— read the case study

Please provide your full name so we know how to address you.

Tell us which company you represent so we can personalise our response.

Use your work email. One of our experts will get back to you.

Which desks would you like to discuss?*

Choose the desks you’d like us to cover. Pick “Not sure yet” if you’d rather we worked it out on the call.

Roughly, so we bring the right person to the call.

Complete the verification to submit the form.