Obin Builds AI for Precision in Financial Operations

“In financial services, when workflows involve capital decisions, 95% correct is 100% wrong,” Apoorv Saxena, CEO and co-founder of Obin AI, told PYMNTS in an interview.

That standard defines how banks, insurers and asset managers approach underwriting, asset servicing, claims management and portfolio monitoring. These workflows directly determine how capital is allocated and risk is priced. Saxena said most existing AI tools do not meet that bar because they optimize for productivity rather than decision quality.

“Copilots are helpful, but they’re about individual productivity,” he said. “If you want agents to take over the workflow, where humans are just supervising, that requires a completely different approach.”

Obin is building what it calls an agentic workforce to execute these workflows end to end. The system operates across banking, insurance and asset management functions, allowing humans to supervise rather than manage each step.

The company frames the opportunity as a capacity constraint. “Financial services has always been constrained by capacity. It’s not an efficiency play,” Saxena said. “If you hire 100 more people, you will have 100 more deals to do.”

That constraint is visible in portfolio coverage. “A typical fund has about 750 companies on its books … they might actually review only four or five,” Valliappa Lakshmanan, Obin co-founder and CTO, said. “What about the other 746?”

Obin’s agents aim to expand that coverage across underwriting pipelines, portfolios and compliance workflows. The goal is to help institutions evaluate more assets more frequently without scaling headcount in parallel.

Lakshmanan said the shift requires moving beyond incremental automation. “Most people build copilots and chatbots, but those are incremental. You still need the human at every step,” he said. “We build these autonomous agents that solve the workflow end to end.”

Saxena positioned the model as additive rather than replacement. “What if we are building AI workers for you?” he said. “If you had 1,000 more associates working for you, what are the things you would do that you’re not doing today?”

As financial institutions push AI deeper into core operations, Obin is focusing on workflows where precision is critical, and scale is limited by human capacity. The company is betting that systems designed to meet the accuracy requirements of capital decisions will define the next phase of enterprise AI adoption.

In other funding news, AI infrastructure and security dominated startup funding this week, with investors doubling down on systems that manage, secure and operationalize autonomous agents across enterprises:

• Oasis Security led the cycle with a $120 million Series B backed by Craft Ventures, Sequoia and Accel, bringing total funding to roughly $190 million, according to Bloomberg. The company focuses on securing “non-human identities,” including AI agents, as machine accounts rapidly outnumber human users in enterprise systems. The bet reflects a growing concern that as companies deploy AI agents at scale, access control and identity governance become core infrastructure rather than peripheral security layers.
• Security remained a central theme with XBOW, which raised $120 million at a valuation above $1 billion to expand its autonomous offensive security platform, per a Bloomberg report. The company builds AI systems that simulate real-world cyberattacks, signaling a shift toward automated, continuous penetration testing as threats scale faster than human teams.
• Beyond security, startups are targeting enterprise workflow automation. Standard Template Labs launched with a $49 million seed round led by ICONIQ and CRV to rebuild IT service management as an AI-native system that resolves requests end to end rather than routing tickets. The platform uses a “digital twin” of enterprise systems to automate coordination across fragmented tools and teams.
• A similar theme appears in Edra that raised $30 million from Sequoia, 8VC and A*. The company applies AI to forward-deployed engineering, using enterprise data to “reverse engineer” workflows and train agents tailored to specific organizations.