Getting Started

Step 1 — Scoping: success metrics defined up front

Before any work starts, we agree in writing on what success looks like in your numbers — captured leads, kept appointments, answered calls, recovered invoices — and what "done" means for the engagement. We don't measure ourselves by tokens burned or hours of "AI activity." This is the same principle stated on Trust & Operating Principles: your business growing is the success metric. If we can't define a metric you'd care about, we'll tell you the service isn't a fit yet.

Step 2 — The pilot: $500 flat fee, two weeks

Instead of a monthly commitment, you start with a fixed $500 flat fee for a two-week pilot, per service — one price stated up front, no hourly billing, no scope creep. During the pilot we deploy a scoped agent against a sample of your real operations and report outcome quality, cost per task, and escalation behavior, so you can see measured performance before deciding whether to scale. The pricing details live on the pricing page and the billing terms in the Terms of Service; if those pages and your signed agreement ever disagree, the signed agreement wins.

Step 3 — What you provide

• Knowledge-base documents — your service policies, pricing, hours, and FAQs, loaded into your workspace's knowledge base. The agents work from your material, not generic scripts.
• Access to the systems in scope — integrations are set up per engagement, for your workspace only, never pooled across clients.
• A named human approver — someone on your side who releases held drafts and answers escalations. Agents score, route, and draft; they are structurally incapable of deciding on their own to contact a person, so approvals are part of the design, not an extra.

Step 4 — During the pilot: guardrails and escalation

Hard rules — booking hours, slot conflicts, spend caps, contact policy — are enforced in code, not in prompts, and every blocked action carries an explicit machine-readable reason. Anything unusual escalates to a human immediately rather than being improvised. You get reports you can check, and every figure from our dry-run simulator is labeled as simulated, not passed off as results.

Step 5 — After the pilot: metered billing you can audit

Ongoing service runs on subscription tiers priced per your signed agreement, with usage metered on a hash-chained ledger — every invoice line traces back to ledger rows of actual activity, so you can verify what you're paying for instead of trusting a summary. The general billing model is described in the Terms of Service; your signed agreement is the specific deal.

Reaching a human, at any step

For anything at all — questions, concerns, a change of mind — email [email protected] and a human answers. If you're an existing customer with a billing, login, or outage issue, open a ticket on the support page; tickets route to the owning team the moment they arrive.

What's honest to expect: this is experimental technology

AI agents are a young technology and parts of this service are new; we'll tell you which of our capabilities are brand-new versus battle-tested rather than pretending everything is mature. The service is provided as-is under the Terms of Service, agents fail sometimes — what we commit to is fast escalation and honest reporting when they do. Where you see performance figures on this site labeled simulated, they are deterministic dry-run estimates, not results from customer engagements.