Answer the repetitive questions without hiring — and without pretending a bot is a person
The job: most of your inbound is the same twenty questions, and every one a human answers is a minute stolen from the case that needed a human. Oneop automates that layer with three modules — the AI chatbot, the help desk with support automation, and live chat for the moment automation should stop. The stopping is designed in, not bolted on.
Sign-up opens when the app launches. We will email you once — no marketing list.
Automation with explicit boundaries, not a bluff
Answers from what you actually know
~55 executable block types, 17 templates, or a flow generated from a plain-English description. Publishing validates the whole graph — a broken flow is refused.
A real handoff, not a buried link
Queue position, an explicit "no agent available" path, a business-hours gate — and the agent inherits the complete transcript through a single choke point.
Routing that survives a restart
Delays that truly pause via a persisted row, approval steps gated by role rank, and a dry-run simulator before you trust a flow with real tickets.
What it looks like in practice
A visitor asks about a return. The flow classifies intent, pulls the policy from the KB, and answers. They ask something the flow does not cover; it offers handoff, and because it is 22:40, the business-hours gate routes to the offline branch and opens a ticket instead.
- What's your refund policy?
- Refunds are handled case by case within 14 days of purchase. Want me to check your order?
- Yes — order 48291
- That order is outside the 14-day window. I'm looping in a person who can make the call — one moment.
- Handed off to Amara · Support
Specific, checkable facts — not a testimonial
No deflection rates or install counts to show you yet — here are mechanisms you can check yourself.
A five-minute sweep escalates SLA breaches idempotently — nothing double-escalates.
A dry-run simulator shows the exact execution trace before a flow touches a real ticket.
AI ticket drafts are review-first and never auto-sent, on every plan.
The modules that do the automating
Free includes 1 chatbot flow — SLA rules are Pro and above
One flat price per workspace, seats included, and AI usage meters as credits on every plan from Free upward.
Extra seats beyond the included count are $5 a seat a month. Full pricing and the plan comparison →
| Plan | Monthly | Seats included |
|---|---|---|
| Free | $0 | 2 |
| Starter | $19 | 3 |
| Pro | $59 | 10 |
| Business | $129 | 25 |
Questions people ask first
How does the chatbot know my content, and can it hallucinate?
Knowledge sources are PDFs you upload, pages you crawl and Q&A pairs you write; the KB-answer block grounds replies in that content. A flow that dead-ends stays dead-ended unless you opt into the AI fallback, so improvisation is a choice you make, not a default. Prompt-injection blocking and PII redaction run on every AI call, on every plan, and cannot be turned off.
What happens outside business hours?
The handoff block takes an optional business-hours gate. Inside hours, the visitor queues for a live agent with a visible queue position and an explicit no-agent-available path. Outside hours, the flow routes to an offline branch you design — typically collecting the question and opening a ticket.
Can I test automation before it touches real tickets?
Yes, twice over. The chatbot editor has an in-editor tester that runs your unsaved draft, and support automation has a dry-run simulator that shows the exact execution trace without writing anything. Publishing a chatbot flow validates the whole graph and refuses a broken one.
Do delayed automations survive a server restart?
Yes. A delay step pauses via a persisted database row and resumes on a one-minute tick with crash recovery. This is the difference between "wait 4 hours, then follow up" actually happening and silently dying with a process.
Can email be automated the same way as chat?
Yes, as an opt-in. Turned on, a support mailbox becomes a ticket pipeline with cross-pipeline deduplication and threading; the same ticket automation rules then apply to those tickets, and agent replies email back on the same thread. Internal notes are never emailed.
What does the AI part cost?
AI usage is metered as per-tenant credits: a chatbot reply weighs 1 credit, and plans include 50 (Free) to 15,000 (Business) per month. All AI runs on Google Gemini — the only provider implemented. With cap enforcement on, the call at the cap is refused rather than billed as overage.
Get started
Automate the twenty questions. Keep a human one click away.
Start free, build your first flow from a template, and run the dry-run simulator before a single customer sees it. No credit card, 2 seats, 1 flow on Free.
Sign-up opens when the app launches. We will email you once — no marketing list.