Spark
A roomier chatbot with custom tone and a real analytics view.
Spark moves beyond simple FAQs with 75 trained intents, a custom tone of voice, and a conversation analytics dashboard your support lead can read on Mondays.
- 30-day refund
- No subscription
- Owned dataset
Everything you get
- Up to 75 FAQ / intent pairs
- 5 knowledge documents
- Choose 1 channel: Website OR WhatsApp
- Custom tone & persona configuration
- Conversation analytics (45 days)
- Branded chat widget colors + avatar
- Email support within 24 hours
From checkout to live bot — what happens next.
Pay once
Secure Authorize.Net checkout, Visa/Mastercard. No subscription.
Submit brief
Tell us about your business + share docs/FAQs in the order page.
We build
Knowledge ingest, retrieval tuning, eval testing — typically 7-14 days.
Go live
Bot deployed on your channels with handoff routing tested.
Other plans
Quickstart
A single-channel chatbot trained on your FAQs and one document.
Growth
A multi-channel knowledge bot with website crawling and human handoff.
Pro
Three channels, advanced analytics, and a tested escalation flow.
Scale
White-glove enterprise rollout with custom integrations.
Still wondering?
If yours isn't covered here, see the full FAQ or reach out.
Talk to usYou get a fully trained, deployed AI chatbot on your chosen channel(s), the underlying training dataset, an analytics dashboard, and a PDF admin guide. Larger plans include human handoff configuration and dedicated specialist time.
Most plans go live in 7–14 business days. Quickstart typically launches within 5 days. Scale (enterprise) takes 3–6 weeks because of the integrations and multilingual setup.
No. We provision and manage the underlying infrastructure as part of the engagement. If you prefer your own keys for compliance reasons, we can BYO on the Scale plan.
Website widget, WhatsApp Business API, Slack, and a REST API for custom integrations (mobile apps, Discord, Microsoft Teams). Channel availability depends on your plan.
PDFs, DOCX, TXT, Markdown, Notion exports, Google Docs, your website (we crawl it), Zendesk macros, or a structured FAQ list. We handle de-duplication and chunking.
We set up a re-ingestion schedule (weekly, monthly, or on-demand) and stamp every chunk with a freshness date. Quarterly retraining is included on Pro and above.
Yes — and it should. Every plan ships with a confidence threshold tuned during evaluation. Below the threshold, the bot offers a handoff instead of inventing an answer.
No. We use the OpenAI API in zero-retention mode (data is not stored or used for model training). Embeddings live in your isolated vector index.
All payments are processed through Authorize.Net using 256-bit SSL encryption. We never store your full card number — Authorize.Net tokenises it at the moment of checkout.
Yes, on the Pro plan and above. We run a redaction layer before the LLM call and substitute redacted values back in the rendered response if needed.