• Kastle raised $24 million in a Series A led by Insight Partners, with Y Combinator, Commerce Ventures and new backer Fifth Wall joining; no valuation was disclosed.
• The pitch is integration over replacement: AI agents that work inside banks' existing core systems rather than requiring a new AI-native stack.
• Kastle's own release cites two different processed-transaction figures — $1.8 billion and $2 billion — a discrepancy the company did not explain.
• The round targets consumer lending operations first, a segment where fragmented legacy workflows and compliance obligations create the manual-labor bottleneck agentic AI vendors are chasing.
• No disclosed valuation and no named bank client leave Kastle's actual commercial traction — not just the agent narrative — unverified.

Kastle closed a $24 million Series A led by Insight Partners on September 17, 2026, to build AI agents that perform operational work inside banks' consumer-lending workflows, according to the company's announcement. Y Combinator and Commerce Ventures, both existing backers, returned; Fifth Wall joined as a new investor alongside unnamed founders and financial-services executives. Kastle disclosed no post-money valuation. The pitch, in the words of co-founder and CEO Rishi Choudhary, is a 'third path' — agents that operate on top of the legacy core systems banks already run, rather than a new AI-native stack requiring migration.

The Pitch: Agents Over Rip-and-Replace

Kastle's core claim is architectural, not just financial: its agents plug into the fragmented, decades-old core systems large lenders already operate, instead of asking a bank to swap in a new AI-native platform. 'We give financial institutions an AI workforce that can operate across the systems they already have,' Choudhary said in the announcement. That framing matters commercially — core migrations at regulated banks routinely run multi-year and multi-million-dollar, and a vendor selling integration over replacement is selling around procurement's single biggest objection: switching cost.

The Cap Table

Insight Partners, a New York growth-equity and venture firm, led the round; Managing Director Rebecca Liu-Doyle said Kastle 'deploys AI agents that can navigate complexity and get high-stakes work done without waiting for multiyear transformation.' Existing backers Y Combinator and Commerce Ventures re-upped, and real-estate-focused VC Fifth Wall joined as a new name — an unusual investor for a bank-operations vendor, and one whose stake here goes unexplained in the release. No dollar breakdown by investor was disclosed, and neither the company nor Insight Partners named a valuation.

A Discrepancy in the Numbers

Kastle's own announcement is internally inconsistent on its one hard traction metric: the release states its agents have processed more than $1.8 billion in transactions, while the company's boilerplate description in the same document puts the figure at $2 billion. Neither number is broken down by volume, dollar value, client count, or time period, and no bank client is named. For a vendor selling itself on operational execution inside regulated lending workflows, an unreconciled $200 million gap in its own headline metric is the kind of detail worth flagging before citing the number at all.

Why Lending Operations First

Kastle is starting with consumer lending rather than deposits or card operations because loan workflows concentrate the manual, cross-system coordination agentic vendors are chasing: underwriting, servicing and collections tasks that route through fragmented legacy cores while carrying regulatory, service-level and risk-management obligations. It is also the most crowded corner of bank-facing AI right now, with a wave of well-funded loan-automation entrants over the past two years. Kastle's differentiation claim rests on working inside existing cores rather than requiring a new stack — whether that holds under a live bank's compliance review is unverified from this release.

What It Signals

The round signals continued investor appetite for agentic-AI vendors selling into banks' operational back office rather than the customer-facing product layer — a segment where the sales cycle is slower but the switching cost, once embedded, is high. It does not yet signal proven scale: no valuation, no named bank client, and an unreconciled processed-transaction figure leave the actual commercial traction unconfirmed. The next verifiable checkpoint is a named enterprise deployment with a disclosed volume, or the round's valuation surfacing in a later filing — until then, $24 million buys Kastle a headcount runway, not a proof point.