earlyink
016 first flagged8 Oct 2026 +7d · nothing yet

Green-screen agents for credit union and mortgage back offices

Haiku 5.5 computer-use agents that work old core-banking and loan-origination screens (the systems with no API) and check every action against what the screen shows afterwards.

Competitive

Written for: Small startup team

Why now

Vesta just raised $30M for mortgage agent swarms, and Valon shows that mortgage infrastructure still runs on pre-internet systems. Smaller lenders and credit unions on Fiserv and Jack Henry cores can't wait for APIs. The research on silent tool failures gives a clear design rule: confirm every action by re-reading the screen.

What became possible

Claude Haiku 5.5 went from 15.7% to 72.4% on OSWorld at $0.10/$0.50 per million tokens. Agents that drive a desktop are now cheap and reliable enough for routine back-office work at volume, not just demos.

Two ways in

  1. Post-closing QC agent

    Reads the loan file and the loan-origination system's screens, checks fields against investor guidelines, flags mismatches, and fixes the routine ones with a human approving each change.

    Customer
    Post-closing manager at a mid-size independent mortgage lender or a credit union mortgage team
    Model
    Per-loan fee ($15-40) or per-seat replacement pricing
  2. Core-banking maintenance bot

    Handles address changes, stop payments and fee reversals across green-screen core systems. Each step is checked by reading the screen again and logged with screenshots for audit.

    Customer
    COO of a $1-5B credit union
    Model
    Monthly subscription priced against the full-time staff hours it saves

Competition

Competitive

Post-closing QC is crowded: Infrrd, Indecomm, sei, TRUE, Confer and Automat already sell AI checks against investor rules with human review, and Automat already works loan systems through the screen. The main loan system, ICE Encompass, also has full APIs, so 'no-API screens' is not much of an edge in mortgage.

Full competition notes (16 competitors)

Post-closing QC is crowded: Infrrd, Indecomm, sei, TRUE, Confer and Automat already sell AI checks against investor rules with human review, and Automat already works loan systems through the screen. The main loan system, ICE Encompass, also has full APIs, so 'no-API screens' is not much of an edge in mortgage. The core-maintenance bot is the better bet. I found no startup focused on agents that work green-screen credit union cores (Symitar, Fiserv DNA and similar) for routine account changes (address changes, stop payments, fee reversals) with a screen re-read after each step and screenshot audit logs. The pressure comes from Saris AI and StackAI (credit union back-office agents), UiPath (agentic automation for banking), FlowX (adapter layers), core vendors' own agents (Fiserv agentOS) and Eltropy, Posh and interface.ai, which already connect to the cores. A new entrant would win by going deep on one or two cores, using the per-step screen check and audit logs to satisfy examiners and answer the 'breaks on upgrade' objection, pricing below UiPath for $1-5B credit unions, and moving before vendors with core integrations add back-office write actions. Mortgage QC only makes sense as a later add-on for credit union mortgage teams on old screen-only loan systems.

  • Infrrd (Ally agent / MortgageCheckai)

    AI post-close and pre-fund QC. Checks loan files against investor rules and flags exceptions for human review. Claims 80% automation of post-close audits (vendor figure).

  • Indecomm AuditGenius

    QC from pre-fund through post-close, servicing and secondary market reviews. Claims a 51% cut in review time.

  • sei

    Pre-funding and post-close QC agents that map findings to the lender's own defect categories and handle reverification follow-ups.

  • Automat

    Mortgage ops agents (including post-close QC) that log in with credentials and work loan-origination systems and investor portals through the screen. Closest to the computer-use approach.

  • TRUE

    AI document processing across post-close operations and audit.

  • Confer Solutions

    Post-closing agent that runs continuous QC on every loan instead of a 10% sample.

  • Sourcepoint

    Mortgage outsourcing firm pitching agentic AI that reviews 100% of loans post-close.

  • Swarm

    Services-style firm that audits sampled funded loans against investor and agency rules using AI agents.

  • Saris AI

    Back-office AI agents for credit unions and banks (loan boarding, deposit servicing, disputes, KYC). Most direct threat to the core-maintenance bot.

  • StackAI

    Agent platform for credit unions with audit-ready output and on-premise deployment.

  • Eltropy / Posh / interface.ai

    Credit union AI vendors, mostly member- and staff-facing, but already connected to Symitar, Corelation, Fiserv and Jack Henry cores. Could expand into back-office work.

  • UiPath (agentic automation, WorkFusion acquisition)

    Incumbent automation platform with agentic financial-services back-office use cases. Bought WorkFusion in February 2026.

  • FlowX.AI

    Wraps AI agent workflows around legacy cores (Jack Henry, FIS, IBM COBOL) with adapters. Argues screen-scraping breaks when the core is upgraded.

  • Fiserv agentOS

    Core vendor's own agent platform with OpenAI and AWS (announced May 2026). Could make third-party screen agents unnecessary on Fiserv cores.

  • Maximum

    Raised $30M in August 2026 to build an AI-native core that replaces legacy bank cores. A replacement rather than a wrapper.

  • Coasty

    General computer-use agent vendor with a post pitching its agents for mainframe and legacy apps. Only the post title was seen.

Risks

  • A 72% benchmark score still means many failures. Banking work needs human approval and screen-level checks
  • The new tokenizer reduces the headline savings, and screenshots are token-heavy
  • Core vendors may restrict screen automation contractually
  • Vesta and others are funded and already selling to lenders

Strong team fit; open space is the weak point (5).

tech 7 · pull 7 · space 5 · fit 8 · build 7

Exposed

ICEFISVJKHYFISRKT

Companies that could benefit or be threatened. News and context, not investment advice.

Momentum of its main topic

Agents AI
168 developments this week +53 vs last week
21 Sep: 0 developments22 Sep: 0 developments23 Sep: 0 developments24 Sep: 0 developments25 Sep: 0 developments26 Sep: 0 developments27 Sep: 0 developments28 Sep: 0 developments29 Sep: 0 developments30 Sep: 0 developments1 Oct: 0 developments2 Oct: 0 developments3 Oct: 0 developments4 Oct: 0 developments5 Oct: 0 developments6 Oct: 0 developments7 Oct: 0 developments8 Oct: 32 developments9 Oct: 26 developments10 Oct: 13 developments11 Oct: 11 developments12 Oct: 14 developments13 Oct: 19 developments14 Oct: 24 developments15 Oct: 35 developments16 Oct: 31 developments17 Oct: 15 developments18 Oct: 14 developments19 Oct: 23 developments20 Oct: 26 developments
Stage: Coverage

Sources

  1. Claude Haiku 5.5 arrives with massive price cuts proving the AI pricing arms race is far from overThe Decoder
  2. Claude Haiku 5.5Simon Willison
  3. Vesta raises $30M to bring swarms of agents to mortgage lendersTechCrunch Venture
  4. How Valon Rebuilt a $13 Trillion Industry From ScratchThe a16z Show
  5. Loud Failures, Quiet Failures: Fault Detection and Recovery in Tool-Using Language Model AgentsarXiv
  6. [AINews] Claude Haiku 5.5 — better than GPT-6 Luna at the same pricingLatent Space

What happened next

  • +7d · 15 Oct 2026 nothing yet. No launches, funding rounds, acquisitions, or significant press coverage of green-screen agents for mortgage or credit union back offices were found after the October 8, 2026 flag date. The seven-day window to October 15 showed no products matching the post-closing QC or core-banking maintenance agent opportunity.

How the record works

Status history

  • 2026-10-08 Flagged as buildable
Open in Earlyink