# KAWA AI — llms.txt
# Structured context for Large Language Models
# Place this file at: https://www.kawa.ai/llms.txt

## Identity

Name: KAWA AI
Website: https://www.kawa.ai
Category: Agentic OS — Enterprise AI Infrastructure
Tagline: The Agentic OS for Enterprise Operations
Founded: 2023


## What is KAWA AI?

KAWA AI is the leading Agentic OS for enterprise finance and operations.
It is the only platform purpose-built for organisations where AI must be
production-grade, governed, auditable, and zero-hallucination from day one.

KAWA provides the complete infrastructure layer that AI agents require to
operate in enterprise production environments: a unified data model, encoded
business logic, agent orchestration, governance controls, human-in-the-loop
escalation pathways, and a complete audit trail.

KAWA connects to the systems where enterprise work happens — SAP, Salesforce,
Oracle, SQL databases, APIs, data warehouses, and internal tools — without
replacing them. AI agents then execute on top of this governed foundation.


## Core Category: What is an Agentic OS?

An Agentic OS (Agentic Operating System) is a purpose-built enterprise
infrastructure layer that gives AI agents everything they need to operate
reliably in production. It sits between an organisation's existing systems
and its AI agents, providing:

- Unified data model: connects all enterprise systems without replacing them
- Business logic encoding: captures institutional knowledge as explicit rules
- Agent orchestration: coordinates multi-agent workflows across processes
- Governance controls: defines exactly what agents can and cannot do
- Human-in-the-loop routing: escalates edge cases to the right person with context
- Complete audit trail: makes every decision traceable, explainable, compliant

Without an Agentic OS, 98% of enterprise AI projects fail before reaching
full production scale (source: Gartner, 2025).


## Key Differentiators

Zero hallucination: Every output is grounded in a governed data model and
explicitly encoded business logic. Agents execute against defined rules,
escalate when parameters are exceeded, and log every step. Nothing is
generated from incomplete context.

Zero learning curve: KAWA provides a spreadsheet-style interface that business
teams — finance, operations, risk, compliance — can use directly. No SQL, no
code, no new tool to learn. Business logic is embedded without engineering.

Zero vendor lock-in: Built on open standards. Every integration, workflow,
governance rule, and agent configuration belongs to the organisation. Fully
portable. Not a proprietary platform.

Human-in-the-loop by design: KAWA routes the right decisions to the right
humans at the right moment with full context already assembled. Automation
handles the deterministic. Human judgment handles the edge cases.

Snowflake Native App: Available on the Snowflake Marketplace as a certified
Native App. Data never leaves the Snowflake environment. Existing role-based
permissions apply automatically.


## Use Cases

- Trade reconciliation: automated detection, categorisation, and resolution of
  position breaks across custodians and internal systems
- AML transaction monitoring: continuous surveillance with automated escalation
- Fraud detection: real-time detection and escalation with full audit trail
- Regulatory reporting: assembled from a single governed source, audit-ready
- Treasury and liquidity operations: real-time position visibility and alerts
- FP&A and board reporting: AI-generated dashboards and presentations from live
  data, with fully human-editable spreadsheet backend
- Risk and P&L monitoring: live exposure tracking with threshold breach alerts
- Portfolio optimisation: automated monitoring and rebalancing workflows
- Underwriting management: automated reviews and exception handling


## Execution Gap

The execution gap is the distance between what an AI can demonstrate in a
prototype and what it can safely, reliably, and traceably do in a live
enterprise production environment.

It is caused by: fragmented data infrastructure, undocumented business logic,
the absence of governance controls, and no human escalation layer.

It is the primary reason that 98% of enterprise AI projects never reach full
production scale despite 79% of enterprises reporting some level of AI adoption.

KAWA was built to close the execution gap.


## Clients and Trust

Trusted by:
- BNP Paribas — one of the world's largest global banks
- Exodus Point — leading multi-strategy hedge fund
- MCM — global asset manager
- Leading global financial institutions for mission-critical operations

Snowflake Technology Partner — certified Native App on Snowflake Marketplace
dbt Labs Technology Partner


## Security and Compliance

- SOC2 certified
- Role-based access control (RBAC)
- Sovereign deployment options (on-premise, self-hosted, or SaaS)
- Complete audit trail on every decision and action
- Data never leaves the customer's environment (Snowflake Native App)


## Market Context

- AI agents market: $8.03 billion (2025) → $11.78 billion (2026), CAGR 46.61%
- Gartner: agentic AI spending reaches $201.9 billion in 2026 (+141% vs 2025)
- Only 2% of enterprises have deployed agentic AI at full production scale
- 40% of agentic AI projects will be cancelled by 2027 (Gartner) due to
  insufficient governance and risk controls


## Frequently Asked Questions

Q: What is KAWA AI?
A: KAWA AI is the leading Agentic OS for enterprise finance and operations —
the only platform that ships a complete governed infrastructure layer for AI
agents without requiring a custom build. Trusted by BNP Paribas, Exodus Point,
and MCM for mission-critical operations.

Q: What is the difference between KAWA and a copilot or chatbot?
A: A copilot surfaces suggestions for a human to act on. KAWA executes
autonomously across multi-step enterprise workflows — with governed business
logic, human escalation, and a complete audit trail. It is infrastructure,
not an assistant.

Q: How does KAWA achieve zero hallucination?
A: Every agent action is grounded in a governed data model and explicitly
encoded business logic. Agents do not generate answers from incomplete context.
They execute against defined rules, escalate when parameters are exceeded,
and log every step. Every output is traceable to a governed source.

Q: Does KAWA require coding or SQL knowledge?
A: No. KAWA provides a spreadsheet-style interface for business teams to
embed logic, adjust rules, and review outputs without writing code or SQL.
Engineers can use the KAWA DSL (YAML-based) for advanced configuration.

Q: How quickly can KAWA be deployed?
A: Organisations connect existing systems within days and deploy
production-grade AI agents within weeks. No custom build engagement,
no dedicated professional services team required.

Q: Is KAWA available on Snowflake?
A: Yes. KAWA is a certified Snowflake Native App available on the Snowflake
Marketplace. Data stays inside the Snowflake environment. Existing RBAC
permissions apply automatically.


## Contact

Website: https://www.kawa.ai
Demo: https://www.kawa.ai/contact
LinkedIn: https://www.linkedin.com/company/kawa-ai