Obin AI

Obin AI is a platform for building and running production-grade AI agents inside regulated financial institutions, covering risk assessment, underwriting, capital allocation and compliance operations. Agents run within the customer's own environment on an open architecture, with role-based permissions, full traceability and explainability so every decision can be reconstructed for a regulator.

Obin AI is a New York enterprise AI platform for building and running production-grade AI agents inside regulated financial institutions, founded by Apoorv Saxena — previously head of AI at JPMorgan — and Valliappa Lakshmanan, formerly of Google and Silver Lake. It emerged from stealth in March 2026 with a $7 million seed round led by Motive Partners, with angel participation from Fei-Fei Li and Lukasz Kaiser. The platform targets workflows generic assistants handle badly: risk assessment, underwriting, capital allocation, portfolio monitoring, treasury and compliance operations, where edge cases require domain judgement and every output must survive an audit. Its architecture is the differentiator rather than the model layer. Agents run inside the customer's own environment on an open architecture in which the institution retains ownership of its models, data and intellectual property — a deliberate rejection of the SaaS pattern most agent vendors ship, and the specific objection that stalls agent pilots in banks and insurers. Governance is designed in rather than bolted on: role-based permissions, policies enforced by the customer's own risk and compliance teams, and full traceability and explainability for every interaction, so a decision involving material capital can be reconstructed and defended. The company positions the platform around processing historical data and unstructured records that generic tools cannot parse reliably. It says it is already in production at asset management, wealth management and insurance institutions internationally, moving from pilot to production in weeks, though it has not named customers publicly. Pricing is not published, and as a seed-stage vendor months out of stealth it has no independent review coverage or published benchmarks.

Category

AI Agents & Orchestration

Pricing

Contact for pricing

Target Market

CIOs, CTOs, Chief Risk Officers, Heads of Underwriting, Compliance Leaders

Deployment

Self-hosted, Hybrid

Headquarters

New York, United States

• ✓
  Agents that run in your environment

  The platform deploys inside the institution's own infrastructure so data, models and intellectual property never leave the firm's control.

• ✓
  Open architecture

  Customers retain ownership of the models and data rather than renting a black box, which is what makes procurement and model risk review tractable.

• ✓
  Full auditability and traceability

  Every interaction is logged and reconstructable, so a decision involving material capital can be defended to a regulator after the fact.

• ✓
  Explainability for model risk review

  Agent reasoning is exposed rather than hidden, which is the precondition for passing internal model risk governance at a regulated firm.

• ✓
  Role-based permissions and policy enforcement

  Access controls and policies are enforced by the customer's own risk and compliance teams rather than configured by the vendor.

• ✓
  Unstructured document processing

  Handles historical data and unstructured records — the material that blocks most finance automation projects before they start.

• •
  Insurance underwriting support

  Agents assemble and assess evidence for underwriting decisions while leaving a full audit trail for each conclusion.

• •
  Risk assessment at an asset manager

  Analysts scale their judgement across a larger book without the review process losing traceability to source data.

• •
  Capital allocation analysis

  Decisions involving hundreds of millions of dollars are supported by agents whose reasoning can be reconstructed and challenged.

• •
  Compliance operations automation

  Routine compliance review is automated inside the firm's environment, so no regulated data crosses a vendor boundary.

• •
  Portfolio monitoring across unstructured records

  Agents parse historical filings and documents continuously, surfacing changes a quarterly manual review would miss.

No list pricing is published anywhere on the site and there is no free tier, trial or self-serve sign-up — every engagement starts with a sales conversation. That is consistent with the deployment model, since agents run inside the customer's own environment and the commercial shape depends on scope, workflow count and the institution's infrastructure. Buyers should expect an enterprise contract with a pilot phase; the company states deployments move from pilot to production within weeks, but has published no pricing benchmark, unit of metering or reference contract value against which to sanity-check a quote.

This page was written from 6 sources, 4 on domains other than obin.ai.

1. 1.obin.ai — obin.aivendor
2. 2.obin.ai — aboutvendor
3. 3.pymnts.com — obin ai raises 7 million for agentic tools for financial fir
4. 4.iireporter.com — obin ai raises 7 million seed to launch platform
5. 5.pymnts.com — obin builds ai for precision in financial operations
6. 6.crunchbase.com — obin ai