Frequently Asked Questions

It runs a continuous loop on one operational surface: plan → act → observe → improve. Concretely: our Voice Support Agent answers inbound calls and books follow-ups; Email Operations triages your shared inbox and drafts replies for approval; Scheduling handles bookings with conflict protection; Lead Qualification scores and routes new leads; Reporting turns raw metrics into a weekly summary with anomaly flags. Each deployment is scoped to one job, with everything else routed to humans.

No. Outbound contact is gated in code, not just in policy: agents cannot send outreach messages — the action is blocked at the platform layer regardless of what the model decides. Email replies start as drafts a human approves. Voice agents answer inbound calls; they don't place outbound ones unless you explicitly enable it later.

Four layers: (1) replies are grounded in your knowledge base and cite their source document; (2) risky categories — refunds, cancellations, anything contractual — permanently escalate to a human via a fixed phrase list; (3) every action is written to an audit log you can review; (4) every human correction becomes training signal through a nightly improvement loop. See our trust and operating principles page.

It escalates. Unintelligible requests, repeated confusion, angry-customer signals, and any policy-listed topic route to a designated human with full conversation context attached. Escalation precision is a tracked metric — we tune it so your team gets real escalations, not noise.

Each vertical has a published monthly price plus usage component — see the full pricing page for all five services and break-even tables. A scoped $500 pilot is the standard starting point: fixed price, defined success metrics, no long-term commitment. All figures are transparent list prices, not custom quotes.

A pilot runs one agent on one surface for a fixed window (typically 2 weeks) against agreed success metrics — response time, draft acceptance rate, bookings made, leads qualified correctly. You get a daily summary and a final report. If the metrics aren't met, you don't expand. We recommend starting with your highest-volume, lowest-risk category and expanding from evidence.

Only what the job requires: the shared inbox, calendar, lead form feed, or metrics source for that vertical — nothing broader. Your data is scoped per project, never reused across clients, and access is documented in the audit trail. You own your data and the improvement-loop artifacts built from it; offboarding means handing over logs, knowledge-base updates, and loop memory.

No. Day-to-day interaction is reviewing email drafts, taking escalations, and reading the weekly report — normal business work. Setup is ours: scoping, guardrail configuration, knowledge-base grounding, and metric dashboards come with the deployment.

Yes — the live demo walks real ticket scenarios through each vertical, including guardrail rejections and escalations, step by step. There's also an interactive ROI calculator to estimate monthly savings at your volumes, and case studies on the blog. Note: demo and calculator figures are simulated illustrations, not quotes.

Mistakes are expected and engineered for: draft gates catch them before send in email, escalation paths catch them in voice, and every error lands in the audit log with the reason behind the decision. The improvement loop's whole purpose is turning each mistake into a permanent fix — corrections proposed today typically show up as behavior change within days.

In plain terms: it handles one repetitive communication job so your team stops doing it by hand. Depending on which agent you start with, that means answering inbound calls and booking follow-ups, sorting your inbox and drafting replies for you to approve, taking bookings without double-booking anyone, scoring new leads and routing the good ones to a human, or writing a weekly report from your numbers. It works on one surface at a time, follows the rules you set, and hands anything unusual to a person. It does not replace you or your staff — it takes the repetitive front-line work off their plate.

We scope honestly, and some things we'll tell you no on. Our agents cover five jobs: voice support, email operations, scheduling, lead qualification, and reporting. They are not a fit for: judgment calls like refunds, cancellations, or anything contractual (those always go to a human); outbound sales outreach or cold calling — agents only respond to inbound requests, they never contact your customers unprompted; open-ended creative or strategic work; or anything requiring licenses your team holds. If your problem is really "I need more people," an agent won't fix it — but if it's "my inbox/phone/calendar eats hours every day," it probably will.