Kognitos AI Employees Arrive to Run Finance SOPs

Kognitos, the San Jose company building neurosymbolic AI for business automation, announced Kognitos AI Employees on October 21 — a library of 61 pre-built AI roles designed to run finance and banking workflows exactly the way a company's approved procedures say they should. The launch also expands Kognitos Context Graphs with 133 ready-made finance monitors that teams can switch on as continuous controls with a single click.

According to the announcement issued by the company, the new Kognitos AI employees draw on a catalog of 515 standard operating procedures covering work such as accounts payable, reconciliation, the month-end close, and customer onboarding checks. A new builder agent called Quill studies a customer's systems and recent transactions and drafts new procedures in plain English for people to approve.

Why determinism is the selling point

The pitch is aimed at a problem finance leaders know well: most AI agents improvise. As reported by Kognitos, research examining 15 models across 18 months of releases found that while accuracy improved, reliability barely changed — even the most accurate agents produced different results from one run to the next. An agent that handles the same invoice differently each time cannot survive an audit.

Kognitos takes a different architectural path. Unlike a generic agent that improvises, a Kognitos AI employee executes each procedure deterministically, in plain English that an auditor can read, the same way every time. According to the company's AI Employees page, numbers shown in its 100-plus connected apps are computed, not generated, and every run leaves a journal recording each step, every system call, and the exact procedure version that ran.

That framing drew outside endorsement. According to the announcement, Rajeev Madhavan, founder and managing director of Radiant, said the shift changes the economics of growth for finance organizations, letting them expand capacity without expanding headcount at the same pace and redeploying senior staff toward cash management, capital allocation, and strategy.

What finance teams actually get

The model is presented like hiring. A team picks a role, reads its profile — including what it will not do and which other roles it works with — and clicks hire. Onboarding is handled by Quill, which drafts each procedure from the Kognitos library or from how the team already works. Supervision is granular: teams choose action by action what the employee may do on its own and what requires a person first.

Procedures in the 515-item library are split along separation-of-duties lines, a deliberate nod to internal controls. The Context Graph side of the launch turns plain-English questions into monitors that run daily and remember every decision. And with new REST and MCP access, customers can connect their own AI agents into the platform — a notable bridge to the wider agent ecosystem, since Model Context Protocol has become the connective tissue between enterprise agents and the tools they use.

The launch builds on the Context Graph for Finance introduced in August, which paired a finance-aware foundation with specialized agents for accounts payable. Kognitos chief executive Binny Gill has argued that finance cannot run on probabilities, and that a general-purpose model designed to generate the most likely response will hallucinate precisely where finance needs consistency. As reported in the August announcement, the company positions its Context Graph as the organization's actual finance structure — policies, approval paths, and controls — that the AI reasons over instead of guessing against.

The company has gathered industry recognition along the way. Gartner named Kognitos a sample vendor for Context Graphs in its 2026 Hype Cycle for Agentic AI, and it took the top overall spot in the 2026 ISG Buyers Guide for Automation and Orchestration. Founded in 2020 and backed by investors including Khosla Ventures, Prosperity7 Ventures, Engineering Capital, and Wipro Ventures — with a $25 million Series B on the books — the company is betting that the next phase of agent adoption belongs to systems that show their work.

AI Employees and the expanded Context Graphs are available in preview to Kognitos customers. For the agent ecosystem, the launch is another data point in a visible shift: from agents that decide each step at runtime to Kognitos AI employees that follow approved procedures, keep a journal, and hand a person the pen whenever the rules say so. It lands in a week already busy for agentic finance — see the coverage of AI agent payment rails and Ant Group's bet on AI financial agents — and suggests the industry's next moat may be procedural trust, not raw model capability.