AI answers your website can stand behind
Publish approved knowledge, embed a branded widget, and add grounded AI when a visitor question does not match your FAQ exactly — with citations validated server-side, never raw model tokens in the browser.
Growth
Add grounded AI answers when approved content does not match exactly.
Live on your site in minutes, not months
No engineering sprint required. Business owners go from signup to a working widget with approved answers the same day.
Create your workspace
Sign up free, verify your email, and pick a name for your site or brand. No credit card required.
Publish approved answers
Import FAQs, write answers, review, then publish — your widget only shows content you approve.
Paste the widget on your site
Copy one embed snippet, set allowed domains, and your visitors get instant answers from your library.
Point it at your website. Knowledge ready in an afternoon.
Your RAG-governed AI chatbot does not need a manual copy-paste marathon. Give us your site URL, let the crawl build draft answers from your existing pages, then approve what you want live.
Paste your website URL
Enter your homepage or help center address. No sitemap upload, no developer ticket — just the URL your customers already visit.
We read your public pages
The platform reads your site and turns FAQ-style content into draft answers organized for review — not live on the widget until you say so.
Review, approve, and go live
Skim the drafts, edit anything that needs a tweak, publish the ones you trust, and your chatbot answers from that approved library immediately.
Website crawl is included on Scale and above. On any plan you can still import FAQs manually or upload documents — same review-and-publish path.
Start free — import your first answersDrafts never go live until you approve them — your widget stays unchanged.
Every way to launch, control, and improve — without a dev team
Import from your site, files, or a spreadsheet. Embed in minutes. Publish with rollback, honest refusals, and feedback that tells you what to fix next.
Drop PDFs and docs
Upload policies, manuals, and help docs. We turn them into draft FAQ answers you review before publish.
Bulk FAQ import (CSV/JSON)
Import CSV or JSON FAQ lists and run the same approve-and-publish workflow your team already uses.
One snippet, live today
Copy a single embed code, set allowed domains, and your visitors get answers — no custom app build.
Suggested question chips
Pin starter prompts so visitors know what to ask first — Shipping, Refunds, Billing, and more.
One-click rollback
Restore a previous publication instantly if a change needs to be undone — without redeploying your site.
Refuses when it does not know
When evidence is thin, the widget takes a safe no-answer or contact path instead of inventing a reply.
Domain allowlisting
Lock the embed to approved origins so your public widget key only works on sites you trust.
Visitor thumbs feedback
Capture helpful vs miss signals so your team knows which answers to fix next.
Usage dashboard and alerts
Track AI credits in the operator console and get alerts before usage surprises — never shown to visitors.
Widget + console
Two settings, no deploy pipeline. Paste the embed snippet, lock domains, and pin suggested chips for Refunds, Shipping, and Billing.
Start free. Add grounded AI when you are ready.
Launch with unlimited approved answers, then upgrade for generation credits, website crawl, and multi-workspace control — on your schedule.
Governed answers first. Smart AI when you need it.
You get the speed of a modern chat widget with the control of a reviewed knowledge library, plus optional grounded generation and BYOK for growing teams.
Unlimited approved FAQ answers
Exact matches from your published library are served with citations — unlimited and unmetered on every plan.
Grounded AI with citation checks
Generate answers only from retrieved approved content, with server-side citation validation before anything reaches the widget.
Branded widget for your site
Match colors, welcome message, and domain lock so the chat feels native to your brand.
Separate workspaces per site
Run multiple brands or clients with isolated knowledge, widgets, and usage — multi-tenant by design.
Bring your own model key (BYOK)
Keep your provider contracts. The platform handles governance, retrieval, and analytics.
Credits only when you generate
Metering applies to grounded AI generation only. Direct approved answers stay free to run.
Not another generic chatbot
Most widgets optimize for volume. Power AI Services optimizes for answers you can defend — with a clear path from approved content to grounded AI when needed.
Typical website chatbot
- Trains on scraped pages and hopes for the best
- Improvises when it does not know
- Charges per seat or per conversation
- Hard to prove which answer came from which source
Power AI Services
- Answers come from reviewed, published knowledge
- Refuses or falls back when evidence is thin
- Unlimited direct answers on every plan
- Citations tie every reply to approved content
Enterprise-grade retrieval — explained simply
Your visitors see clear answers. Behind the scenes, a deterministic pipeline searches approved content, fuses lexical and vector results, and validates citations before anything reaches the widget.
Typed or tapped in the widget
Only allowlisted domains are served
Keyword and meaning search over published chunks, fused by rank
Is the approved evidence strong enough to answer?
Direct exact answer · Grounded answer + citation check · No-answer + contact path
Sent to the widget after checks — never raw model tokens
Hybrid retrieval (RAG)
Keyword and vector search fused by reciprocal rank fusion over published chunks only.
Published chunks only
Drafts never answer visitors. Thin evidence triggers a safe no-answer path.
Governed ingestion
Crawl, upload, or CSV → drafts → human approve → then embeddings and publish.
Server-side citation validation
Unknown citation IDs are blocked. Visitors only see answers tied to approved content.
Tenant isolation
Schema-per-tenant PostgreSQL with row-level security keeps each customer’s data isolated.
Metering and BYOK
Exact matches stay unlimited. Meter grounded generation only — or bring your own key.
Start free. Pay only when you generate.
Direct approved answers stay unlimited on every plan. AI credits apply only to grounded generative answers — so your earliest path stays free to run.
Unlimited approved answers. Perfect to launch a governed FAQ widget.
Add grounded AI answers when approved content does not match exactly.
Higher generative capacity for busy sites and multi-brand teams.
Bring your own model key. Platform fee only — no AI credit metering.
Questions business owners ask
What is unlimited on the free plan?
Direct approved answers — when a visitor question matches published FAQ content, you serve it with citations and no AI credits are consumed.
When do AI credits apply?
Credits meter grounded generative answers — when the platform synthesizes a reply from retrieved approved content because no exact match exists.
Can I bring my own model provider key?
Yes. Growth and above support BYOK, so you use your existing model contract while governance, retrieval, and usage analytics stay on the platform.
Do visitors see billing or usage limits?
No. Billing, credits, and plan details live in your operator console only. The visitor widget shows answers and your branding — never account internals.
Put governed answers on your site today
Create a free workspace, publish your first approved answers, and embed the widget — then add grounded AI and BYOK when your team is ready.