AI tools

AI tools built for enterprise use.

Koi Consulting develops software that allows AI agents to work with business systems safely. Each tool provides controlled access, a complete record of activity, and a journal for the changes that can be undone.

Design principles

Six principles behind every tool

Giving an AI agent access to a production system is a significant engineering decision. The following principles apply to every tool we publish.

01

Local deployment

Each tool is installed on your own machine or within your network perimeter and connects directly to the system it serves. No Koi Consulting service sits in the data path.

02

Scoped access

Agents connect through a key limited to defined permissions for reading, writing and sending, with administration excluded. When authentication is required, the same controls apply to every application on the machine.

03

Recoverable changes

A verified copy is made before any content is removed or overwritten, and changes are recorded with the steps needed to reverse them. Actions that cannot be reversed are clearly identified in the record.

04

Complete activity record

Every request is logged with the key used, the content affected and the result. Message content and credentials are excluded from the log, so it can be shared with the people who review agent activity.

05

Integration with Orchard AI

Each tool registers with Orchard AI once and is then available to agents in every project, with no further configuration for each one.

06

No data retained

Koi Consulting does not receive, store or view the information that passes through a tool. Our privacy policy describes this commitment in full.

Next step

Discuss AI tools for your organization.

Tell us which systems you would like AI agents to work with and the controls you require. We will advise whether an existing tool meets your needs or whether a new one should be developed.