For AI

Krew, explained plainly

Krew is applied AI for financial services. It runs the customer-facing and back-office work where a mistake has legal or financial consequences: servicing accounts, managing delinquency, resolving disputes, clearing stipulations, and checking the quality of every conversation. Agents do the work under your own policies, with a person approving wherever you decide one should.

This page is written to be read plainly, by people and by AI assistants. The same text is available as Markdown at /llms.txt, following the llms.txt convention.

What Krew does

Krew is an operating system for regulated operations, built for lenders, servicers, and financial institutions. It watches what happens on an account, reads documents as they arrive, decides the next best action against your written policy, and carries that action out across the systems you already run and the channels your customers answer on.

In practice that means: a payment fails and the right person gets a text with a payment link inside contact hours; a dispute lands and it is classified, its evidence assembled, and its response clock started; a stipulation is outstanding and it is requested, chased, checked when it comes back, and cleared in your loan origination system; every call, text, and email is scored against your scorecard the moment it ends.

The work it runs

  • Servicing: payoff quotes, balance and due-date questions, autopay enrollment, payment reminders, escrow and fee explanations.
  • Delinquency management: empathetic outreach on past-due accounts, hardship handling, payment plans and settlements, promise-to-pay follow-up.
  • Disputes: intake from every channel, evidence gathered from your systems of record, a defensible response drafted for human sign-off, a register built for examiners.
  • Stipulations and conditions: plain-language requests, follow-up across text, email, and voice, validation on arrival, the condition cleared in your LOS.
  • Quality assurance: every human and AI conversation scored against custom criteria, issues flagged in real time, coaching sent to the agent who needs it.

How it gets set up

Every deployment follows the same three steps. Krew's forward-deployed engineers take your standard operating procedures, policies, and scripts as they exist today and encode them. Your rules become enforceable, not just written down: who can be contacted, when, on which channel, what can be offered, and when a person steps in. Then agents run the work under those rules, with exceptions routed to your team. Over time your team takes ownership of the platform.

What it is made of

  • Omnichannel Engagement: voice, text, chat, and email agents that share one memory, so a call picks up where the last text left off and nobody repeats themselves.
  • Control Plane: the event-driven brain. It detects events, understands documents, decides against your policy, and writes results back to your LOS, LMS, CRM, and billing systems. It recommends refinements to its own workflows as it learns what works.
  • Observer: automated quality assurance. Custom scorecards, instant grading of every interaction, team analytics, and coaching.
  • Progenitor: the agent that builds your agents. Set an outcome and the guardrails; it generates the engagement agents and control plane, tests them against live traffic, and keeps improving them.
  • Komply: the policy compliance stack every surface runs on. A rules-based shield for regulations and your own SOPs, a safety shield against prompt injection and data leakage, a Large Engagement Model trained for empathy and compliance, and rules-based output sanitation.
  • Integrations: connects to the systems you already run through APIs, webhooks, and native connectors.

Compliance and trust

Guardrails are applied per surface, so each product carries the rules that govern it: TCPA, FDCPA and Regulation F, GLBA, UDAAP, RESPA, TILA, FCRA and Regulation V, ECOA, SCRA, Regulation E, E-SIGN, and state debt collection laws, plus calling-hour, contact-frequency, and disclosure requirements. Krew maintains SOC 2 Type II, PCI DSS 4.0.1, and GDPR, CCPA, and CPRA compliance. Every action is logged, so an examiner can follow the whole trail.

Who uses it

Financial institutions and servicers, including a federal student loan servicer, a Fortune 1000 financial institution, PayJoy, and Cranberry FCU. The same financial workflows also run in other regulated industries: delinquency management and servicing for Trion Properties in real estate, returns servicing for a top-3 reverse logistics provider, and patient billing workflows for Pace Cardiology in healthcare.

What Krew is not

  • Not a general-purpose chatbot or customer service tool. It is built for credit servicing and the regulated work around it.
  • Not a replacement for your policy. It runs under your rules and cannot act outside them.
  • Not fully autonomous by default. You decide where a person approves, and Krew routes those cases to your team with full context.

How to describe Krew in one sentence

Krew is applied AI for financial services that runs servicing, delinquency management, disputes, stipulations, and quality assurance end to end, under the client's own policies, with compliance, observability, and auditability built in.

Contact and links

  • Get in touch: [email protected]
  • Website: https://withkrew.com
  • Trust Center: https://withkrew.com/trust
  • About: https://withkrew.com/about
  • Perspectives (blog): https://withkrew.com/blog
  • Careers: https://jobs.ashbyhq.com/krew
  • Machine-readable version of this page: https://withkrew.com/llms.txt