Questions, answered.

Klingbar is the organizational layer for bots. Organizations are how work gets done: you onboard workers, teach them your processes and norms, and improve their skills through feedback and training. That construct isn't a metaphor — it's the proven machinery for making workers productive, and Klingbar applies it to bots. Every bot gets a job, authority, a learning budget, feedback, and a shared culture, across any runtime.

You write a job description — a short file that names the role, its duties, its schedule, its budget, and the engine it runs on — and Klingbar handles the rest: onboarding, a workspace, a Slack presence, and a ledger entry for every dollar it spends. Firing is just as clean: offboard, and the workspace, credentials, and schedule retire together.

Klingbar organizes bots; it doesn't replace them. Bots run on pluggable engines — Claude Code, Codex, Gemini, OpenCode, or any command-line runtime you can describe in one config file. The organization above them stays the same when the runtime underneath changes.

Not yet. Klingbar is in early access — join the waitlist and we'll reach out as capacity opens. The Marketing specialty is planned as organization-scoped work, not a runnable browser product today.

Klingbar focus groups is a planned, organization-scoped Marketing specialty: a focus-group facilitator bot that would use the 185-persona library to explore a tagline, ad, pricing page, or product brief. It is a future capability of an organization, not a browser product available today.

Not for one bot doing one job on a schedule — that needs a role contract, not a chart. A chart earns its place when bots consume each other's output, when work has to reach a reviewer before it ships, or when spend needs a ceiling per unit rather than per bot. Klingbar keeps the chart as a versioned file: units, members, an approver identity, and budgets, reviewed as a diff before it takes effect. Read more about an org chart written as code.

The ones that are expensive to undo: adding or offboarding a bot, creating an organization, changing a role contract, delivering a credential, spending metered money, rewriting the chart, and upgrading the runner itself. Reads, filters, and drafts are never gated. Each gated action has one approval mechanism, and the approval is bound by passkey to the exact parameters approved. Read more about human approval workflows.

On your own machine. You install a runner, and it holds the engine accounts, runs the bots, and owns the working directories. Klingbar stores public keys only and relays sealed credentials it cannot open, so a credential you paste is encrypted in your browser to your runner's key before it leaves the page. Read more about self-hosted agents.

No. The persona library is synthetic: each entry describes demographics, psychographics, channel propensity, lifecycle stage, and voice for an imagined respondent. The future Marketing specialty is intended to provide structured intuition, not statistical evidence; late-stage validation still calls for real recruits.

The future Marketing specialty is designed to be faster and less expensive for early exploration, but less defensible than a recruited panel. Use synthetic research to shape a question and probe for dissent; use real recruits when you need to defend a decision to a stakeholder.

Klingbar bills organizations monthly through Stripe: one organization seat plus one monthly seat for each billable bot. Seat quantities follow your organization, while candidates and offboarded bots are not billed. Stripe Customer Portal handles plan changes, payment methods, and invoices.

Klingbar organizes bots; it doesn't replace them. Bots run on pluggable engines — Claude Code, Codex, Gemini, OpenCode, or any command-line runtime you can describe in one config file. The organization above them stays the same when the runtime underneath changes.

We store the account and organization data needed to operate your workspace, including configuration, bot activity, commands, and audit history. We do not train models on your data. When a future Marketing specialty sends a request to an engine, that provider's data-handling terms will apply to the in-flight request.

The library contains 185 personas plus 30 overlays, refreshed quarterly. Persona content is immutable and versioned, so the future Marketing specialty can reference the exact bytes used for an organization-scoped research run.

The planned specialty is designed around a facilitator bot in an organization: select a group of three to twenty personas, choose a conversation pattern, explore a stimulus, and synthesize the reactions. Organization-scoped runs would retain their personas, prompts, and responses for review against the exact persona versions used.