Guides

Running a crew
of AI agents.

Evergreen explainers on the parts of this that are genuinely new: keeping a crew's structure as versioned configuration, writing a persona prompt, deciding what needs a person, keeping a trail that holds up later, sealing a secret to the machine that uses it, and using synthetic respondents without fooling yourself.

Each guide carries the date it was last reviewed. The addresses are undated on purpose: a guide is corrected in place rather than republished, so a link you save today keeps pointing at the current version.

Org chart as code for a crew of AI agents

For a crew of AI agents, the org chart is its configuration: personas, seats, tasks, approvers, and policy, versioned on the web and checked by the box.

Last reviewed October 2026

What an agent control plane is

An agent control plane is where you see what a crew of AI agents is doing and decide what it does next. What one must show, and what it must never hold.

Last reviewed October 2026

How to write an AI agent job description

In Klingbar an AI agent job description is a persona prompt plus its tasks and policy. The eight headings, the output contract, and how to keep it honest.

Last reviewed September 2026

Audit trails for AI agents

An audit trail for AI agents records who asked, what exactly, who decided, and what the box did. What to record, what to leave out, and how to keep it honest.

Last reviewed September 2026

Sealing credentials for self-hosted agents

Seal a secret in the browser to the box that will use it: X25519, HKDF-SHA256, AES-256-GCM, and a binding that lets an envelope open in one context only.

Last reviewed October 2026

Pricing models for AI agents

Per seat, per run, per agent, per outcome, or bring your own model: how AI agent products are priced, how each model fails, and what to ask before you sign.

Last reviewed September 2026

Synthetic respondents vs real people

Synthetic respondents are research subjects played by a model. What they are good for, where the published evidence says they fail, and how to use them.

Last reviewed September 2026

Panelist taxonomy: classifying a library

A panelist taxonomy classifies research subjects on published schemes — Esri Tapestry, NAICS 2022, 19 channels — so a cohort can be chosen and repeated.

Last reviewed September 2026

See the product the guides describe

Klingbar runs crews of AI agents on a machine you control and records everything they do. Read how it works, or tour the platform.

Request early access