AI support that
doesn’t make things up.
Answers grounded in your company’s documents.
Every supported answer includes its source.
No answer in your docs? Keenyx says so.
There’s no permanent free tier — Starter starts at $15/mo…
Source: Pricing
Sounding confident isn’t the same as being right.
Most AI tools are built to always have an answer, so they fill gaps with plausible-sounding guesses. Keenyx is built the other way round: retrieval decides what the model is allowed to say, so an answer it can’t back up isn’t a risk, it’s not an option.
“Do you offer refunds after 45 days?”
Yes — we offer a 60-day refund window for all purchases, no questions asked.
Invented — not in any policy
Our refund policy covers purchases within 30 days — 45 days falls outside that window. I can flag this for a teammate if you'd like to ask about an exception.
Source: Refund Policy
How it works
Four steps, every replyUnderstand
Your question becomes a vector, a numeric fingerprint of what it actually means.
Retrieve
The most relevant passages in your knowledge base are pulled by similarity, not keyword luck.
Ground
Only those retrieved passages are handed to the model. It can't reach outside them.
Cite
The reply ships with the exact source it came from, every single time.
Built for teams who live in their docs.
SaaS
Answer product questions straight from your help center and changelog.
E-commerce
Handle shipping, returns, and policy questions without a human in the loop.
Professional Services
Turn your SOPs and client documentation into instant, accurate answers.
Internal Knowledge
Point it at your internal wiki so your own team gets faster answers too.
Works with the website you already have.
Keenyx doesn’t require a rebuild. Drop one script tag onto any site, and the widget just works.
<script src="https://keenyx.com/widget.js" data-bot-id="your-bot-id" async></script>
Your data stays yours.
Isolated by design
Row-level security keeps your workspace, bots, and conversations invisible to any other account.
Server-side API keys
Your Gemini key lives on our servers only. It's never sent to the browser or exposed in your embed code.
Encrypted in transit
Every connection, dashboard and widget alike, runs over HTTPS by default.
Rate-limited by default
Per-visitor and per-IP limits keep the service fair and resistant to abuse, out of the box.
Card details never reach us
Payments go straight to Razorpay's checkout. We never see or store your card number.
Delete everything, anytime
One action in Billing permanently removes your workspace, bot, knowledge base, and every conversation.
Pricing.
Starter is live today at half price while we onboard early teams. Growth and Scale are being finalized.
Starter
For a single bot doing one job well.
Growth
For teams running a bot across more of the business.
Scale
For high volume and dedicated support.
What’s included
Every plan cites its sources| Starter | Growth | Scale | |
|---|---|---|---|
| Bots per workspace | 1 | 2 | Custom |
| Team members | 1 | Multi-user | Multi-user |
| Conversations / month | 1,000 | 10,000 | Custom |
| Messages per conversation | 20 | 30 | Custom |
| Knowledge sources | Unlimited | Unlimited | Unlimited |
| Citations on every reply | Yes | Yes | Yes |
| Conversation analytics | — | Yes | Yes |
| Reranked retrieval | — | Yes | Yes |
| Voice agent | — | Coming soon | Coming soon |
| Automation workflows | — | Coming soon | Coming soon |
| Support | Priority | Dedicated |
Billing
questions.
What counts as a conversation?
One visitor session with your bot, however many messages it contains, up to the per-conversation message cap.
What happens if I go over?
The bot keeps answering to the end of the billing period and we flag the overage in Billing. Nothing is cut off mid-conversation.
Can I cancel any time?
Yes. Billing has a single action that ends the plan and, if you want, permanently deletes the workspace, bot, knowledge base and every conversation.
Do you store my card details?
No. Payments go straight to Razorpay's checkout. We never see or store your card number.
Grounded answers today. Voice and automation next.
Voice
SoonThe same grounded answers, now on the phone, for callers who'd rather talk than type.
Automation
SoonConnect Keenyx to your existing tools so a grounded answer can trigger the next step automatically.
Questions, answered.
Every reply is generated only from the knowledge base you upload: PDFs, docs, or a website crawl. If the answer isn't in there, the bot says so instead of guessing, and cites the source it used.
Keenyx runs on Google's Gemini by default. If your team needs a specific model instead, whether that's OpenAI, Anthropic, or another provider, we can build that in for you.
No. Your knowledge base and conversations are used only to answer your own visitors, never to train or fine-tune a model.
Most teams are live in under ten minutes: paste in docs or a URL, copy one script tag onto your site, and the widget is answering questions.
We build custom AI agents around your industry’s workflows.