Kastle
by Kastle AI Inc.
AI voice, SMS and chat agents for mortgage servicing, collections and loan operations at banks and lenders
Kastle builds AI agents for lending operations, starting with mortgage servicing and consumer lending. Its agents handle borrower calls, payment collection, loss mitigation and loan officer support on top of a lender's existing servicing and contact center systems, with every interaction scored against FDCPA, TILA, RESPA, UDAAP and TCPA rules.
Kastle is a San Francisco company founded in 2024 by Rishi Choudhary (CEO), formerly the founding product manager of Redfin's Mortgage Marketplace, and Nitish Poddar (CTO), formerly a tech lead at Verkada. It came out of Y Combinator's Summer 2024 batch with AI voice and SMS agents that collect mortgage payments, answer loan and escrow questions and qualify new inquiries, and it has since grown into a platform for loan operations across voice, SMS, email and chat. The product has an Agent Builder that turns a lender's written SOPs into agents, simulation testing for edge cases before go-live, analytics, and a compliance engine pre-trained on FDCPA, TILA, UDAAP, RESPA and TCPA that scores every interaction, so QA covers 100% of calls rather than a sample. Prebuilt agent roles include a loan officer assistant, collections, servicing, loss mitigation and compliance. The agents run on the lender's existing servicing system (National Mortgage News reports an integration with ICE's servicing platform), contact center and payment processor, and hand off to humans for cases such as loan modification requests. In July 2026 Carrington Mortgage Services, which held over $200 billion in unpaid balances, chose Kastle to deploy autonomous and agent-assist AI across its servicing contact centers. Kastle says it is in production at 10 of the top 25 US servicers. In September 2026 it raised a $24 million Series A led by Insight Partners to expand from servicing into bank consumer lending.
The head of loan servicing or contact center operations at a mortgage servicer or bank, who carries high call volume, collections targets and a heavy compliance QA load.
Borrower calls and collections handled around the clock by agents, with every interaction checked against lending regulations.
At a Glance
- Category
- Industry & Government
- Pricing
- Contact for pricing
- Target Market
- Mortgage Servicers, Banks, Consumer Lenders, Heads of Loan Servicing
- Deployment
- Cloud-only, API-based
- Founded
- 2024
- Headquarters
- San Francisco, United States
- Team Size
- 11-50
- Customers
- Claims production use at 10 of the top 25 US mortgage servicers (vendor claim, Oct 2026)
Key Features
- ✓Agent Builder from SOPs
Operations teams describe procedures in natural language and Kastle turns them into agents, so building one needs no engineering ticket.
- ✓Compliance Engine
Pre-trained on FDCPA, TILA, UDAAP, RESPA and TCPA, it scores every call and message, which replaces sampled QA with full coverage.
- ✓Simulation testing
Agents are run against edge-case scenarios before production, which cuts the risk of a non-compliant collections call reaching a borrower.
- ✓Omnichannel, multilingual agents
The same agents work over voice, SMS, email and chat in several languages, matching how borrowers actually contact a servicer.
- ✓Prebuilt lending agent roles
Loan officer assistant, collections, servicing, loss mitigation and compliance agents ship as starting points for the most common loan workflows.
- ✓Human handoff and systems-of-record updates
Agents leave loan comments, update the servicing system and route judgment calls such as modification requests to staff.
Capabilities
Use Cases
- •Inbound servicing calls
Agents take payments, answer escrow and payoff questions and log loan comments at any hour, so borrowers stop waiting in queues.
- •Early delinquency outreach
Collections agents send payment reminders and offer repayment plans to past-due borrowers, escalating hardship and modification cases to people.
- •Full-coverage compliance QA
Every AI and human interaction is scored against lending regulations, giving compliance teams complete coverage instead of a small sample.
- •Loan officer assistance in originations
An agent qualifies inbound inquiries, books appointments, generates pre-approval letters and transfers warm leads to loan officers.
Ideal For
Best For
- ✓Mortgage servicers with volatile inbound call volume around payments and escrow
- ✓Early-stage delinquency outreach and past-due collections with repayment plan options
- ✓Moving compliance QA from sampled calls to full coverage under FDCPA, TILA, RESPA, UDAAP and TCPA
- ✓Lenders that want AI on top of their current servicing system and telephony without replacing either
Not Ideal For
- ✗Lenders outside US consumer and mortgage lending, since the compliance engine and agent roles are built around US lending regulation
- ✗Teams that want a general-purpose agent builder for many departments; Kastle is a vertical product for loan operations
- ✗Buyers who need published pricing or independent user reviews before a demo, since neither exists yet
Deployment
Market & Ratings
Claims production use at 10 of the top 25 US mortgage servicers (vendor claim, Oct 2026)
Market Analysis
Pros
- ✓A named large deployment: Carrington Mortgage Services, with over $200B in unpaid balances, chose Kastle in July 2026
- ✓Compliance scoring on every interaction addresses the main regulatory risk of automating collections
- ✓Works on existing servicing and contact center systems, including ICE's servicing platform, so adoption needs no core replacement
- ✓SOC 2 Type II and a public trust page
Cons
- ✗Young company (founded 2024, about 15 staff on its YC profile) selling into conservative, heavily regulated banks
- ✗Headline results (70% lower cost per call, $1.8B to $2B in transactions processed) come from the company and are not independently audited
- ✗National Mortgage News noted that the Carrington announcement says nothing on borrower consent or disclosure for AI-handled collections
- ✗No G2, Capterra or Hacker News reviews, so there is no independent practitioner feedback yet
Pricing
Enterprise
Contact for pricing
- ✓Agent Builder and simulation testing
- ✓Compliance Engine with full QA coverage
- ✓Voice, SMS, email and chat agents
- ✓Servicing, contact center and payment integrations
Kastle publishes no prices; deployments are sold through demos to servicers and banks. Its own site claims 70% lower cost per call and 40% lower handle time, which are vendor figures, and contract values for named deals such as Carrington's were not disclosed.
Security & Compliance
Connect
Sources
This page was written from 6 sources, 5 on domains other than kastle.ai.
- 1.kastle.ai — kastle.aivendor
- 2.pulse2.com — kastle raises 24 million series a to build ai workforce for
- 3.ycombinator.com — kastle
- 4.nationalmortgagenews.com — carringtons 200b in servicing goes ai native with kastle dea
- 5.beinsure.com — kastle raises 24 mn series a for ai lending workforce
- 6.fondo.com — kastle launches
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
Vesta
AI-native mortgage loan origination system where staff and AI agents work the same loan file
Atomic
AI-native supply chain planning that simulates every SKU daily and issues the purchase orders your ERP executes
Outmarket AI
AI workflows for insurance brokerages: quote comparison, policy checking, proposals and certificates wired into your agency management system
Autonomize AI
Healthcare agent platform and Context AI layer for prior authorization, utilization management and payment integrity