Ship an AI chatbot without touching an API key.
Say “add a support chatbot that knows my services and opening hours” and opix builds the whole thing into your app: the chat UI, the streaming backend, and a metered AI endpoint with a budget cap you set. There is no key to create, paste, rotate — or leak.
Free welcome credits · no card required · failed builds cost $0
The chat UI was never the hard part
Anyone can render chat bubbles. The hard parts of shipping a chatbot are everything around them: provisioning an API key and keeping it out of the browser, streaming responses without a hanging spinner, failing honestly when the model is down, and — the one that hurts — capping what it can cost you when a stranger on the internet decides to chat all night.
opix treats those as the product. When you ask for a chatbot, it generates the UI and wires it to the opix AI gateway: a metered endpoint that runs server-side, bills transparently through your credits, and refuses to exceed the budget cap you set for the app. The key-management problem is solved by there being no key to manage.
What the gateway gives you
The whole chatbot, generated
Chat UI, streaming responses, conversation state and the system prompt — built into your app’s code where you can read and change it.
Hard budget caps
Set a monthly AI budget per app. The gateway enforces it server-side — the worst case is a polite “try again later”, never a four-digit invoice.
No keys to leak
Nothing to paste into client code, commit by accident, or rotate at 2 a.m. The endpoint is metered and authenticated on opix’s side.
Metered like electricity
Every conversation’s cost lands on the same live meter as your builds. You always know what your chatbot costs to run.
Honest failure modes
Over budget or provider down, the bot degrades visibly — a designed notice, not a fake answer or a silently dead widget.
Beyond the widget
Wire the same AI into Telegram or Slack through connectors, or add AI features — summaries, drafting, classification — anywhere in your app.
From idea to answering
Describe the bot
What it should know, how it should sound, what it must never do. opix writes the system prompt into your code, where you can refine it later.
Set the cap
Choose the monthly budget the gateway may spend for this app. Change it anytime — it’s enforced server-side either way.
Publish
The bot ships with your app. Watch conversations and their cost on the meter, and tune the behaviour by describing the change.
What it costs — precisely
Chatbot usage is metered through the same credits as builds, at the model’s real cost plus a small margin, visible on your usage page. A support bot answering a few hundred questions a month typically consumes a small fraction of a plan’s allowance — and the cap you set is the ceiling either way.
Compare that with the standard route: an API key with an open-ended bill attached to your card, plus the engineering time to stream, secure and meter it yourself. On opix, the cap is the feature. Everything else is the feature working.
Asked before you asked.
Which AI models power the chatbot?+
The opix AI gateway routes to current production-grade models and is maintained on opix’s side, so improvements arrive without a rewrite. Your app’s code talks to one stable endpoint rather than a vendor SDK.
Can I use my own OpenAI or Anthropic key?+
Not inside opix — the gateway is deliberately keyless there. If you eject your app, the AI endpoint is a normal server route in your code that a developer can point at your own provider account with your own key.
Can the bot answer from my content?+
Yes — give it your services, FAQs and policies when you describe it, and opix builds that knowledge into the system prompt. For deeper needs, describe them (“answer from my price list, escalate to email otherwise”) and opix generates the behaviour into the app.
What stops abuse from running up my bill?+
The budget cap bounds the worst case and is enforced server-side, and usage is metered per app, so a spike is visible on your usage page immediately — with the cap as the hard stop.
What happens when the budget runs out?+
The bot pauses with a designed notice instead of failing silently, and your app keeps working. Raise the cap or wait for the next cycle — your choice, one click.
Does building or fixing the chatbot cost extra?+
Building it costs normal build credits, shown on the meter like any other change. Failed builds and auto-fixes are free — the $0-failure policy applies to chatbots like everything else.
Your app answers questions by Friday.
Describe the bot, set the cap, publish. Welcome credits are free — no card, no API keys.