AI with an admin panel, not a black box.

BIG AI is the mortgage industry’s first open, admin-controlled AI suite, built by Mortgage Automation Technologies. It runs in an isolated AWS Bedrock environment for each lender, is configured by the lender’s own admins (prompts, agents, data scope, guardrails), logs every answer for compliance, and is grounded in the live Encompass loan file through BIG POS.

$1,000 per month for unlimited users across the whole company — loan officers, processors, underwriters and admins — plus a $2,500 environment setup. No per-seat fees. The only variable is model token usage, passed through at our AWS Bedrock bulk rate: a W-2 borrower file (pay stubs, W-2s, ID, bank statement) runs about $0.10; a heavy self-employed file with two 300-page tax returns and twelve months of bank statements about $2. A lender closing roughly $180 million a month reviews about 4,000 documents for around $300 — so the whole bill lands near $1,300. Because MAT buys tokens in bulk for every client through AWS, usage runs 30–40% below going to a model vendor directly.

Any model on Amazon Bedrock — Anthropic Claude, Amazon Nova, Meta Llama, Mistral and the rest of the Bedrock catalog — can be enabled inside your own isolated environment. Admins choose which models each agent uses; nothing is shared with other lenders.

Each lender gets a dedicated, isolated AWS Bedrock environment — there is no cross-client learning; nothing another lender does ever touches your AI. And because BIG AI runs on AWS Bedrock, Amazon’s own guarantee applies: client usage is never used to train AI models. Admins control what the AI can see field by field, and every interaction is logged for audit.

Document review (pay stubs vs. 1003, business tax returns, self-employed cash flow), purchase-contract extraction, guideline and eligibility questions, in-depth loan-file review and planned pre-closing compliance checks — all inside BIG POS.