An assistant that acts, not just answers
TalkOmni's assistant looks things up and does things in your own systems: captures a lead, books an appointment, checks an order, opens a ticket. The built-in actions work with no code, and your own actions run over a single signed endpoint you control.
From answering to acting in three steps
- 1
Turn on the actions you want
Your plan comes with a set of actions. Enable the ones this project should use from the Tools tab; anything your plan does not include can be added as an add-on.
- 2
Point us at your backend
For actions that touch your own systems, set one HTTPS base URL and a secret. Every call arrives as a POST to that URL, signed with HMAC-SHA256, so you can verify it really came from us.
- 3
Let it work
The assistant picks the right action from the conversation, fills its parameters and calls it. Reads run freely; anything that writes is whitelisted and validated on our side first.
What the actions layer gives you
Built-in actions, no code
Lead capture, appointment booking and business info run inside TalkOmni. You build nothing: the results land in your panel and go out on webhooks.
One endpoint for all your actions
Actions that touch your systems ride a single gateway: one base URL, one secret, no per-action credentials to manage. Every request is HMAC-signed and verifiable.
Define your own actions
Give an action a name, describe in plain language when the assistant should use it, and declare its parameters. It runs over the same gateway, so there is no new URL or secret.
Writes are gated, not trusted
Reads are free; anything that changes your data is whitelisted and its parameters are validated server-side. The model carries intent, never prices or ids it should not decide.
You see every call
Each call to your backend is listed with its request, response, status and duration, next to the conversation that triggered it.
Same actions on every channel
WhatsApp, Instagram, Messenger, Telegram, email, your website widget and phone calls all use the same action set. You configure it once, per project.
Frequently asked questions
- Do I have to build anything?
- Not for the built-in actions: lead capture, appointment booking and business info run inside TalkOmni. You only build when an action needs data that lives in your own system, and then it is one HTTPS endpoint, not one per action.
- How do you keep it safe?
- Reads run freely; anything that writes is whitelisted and its parameters are validated on our side before the call. Every request to your backend carries an HMAC-SHA256 signature over the timestamp and the raw body, so you can reject anything that is not from us, and every call is logged.
- Which actions do I get?
- Your plan includes a set of general actions, and higher plans include more. Sector-specific action packs and extra custom-action slots are sold as add-ons, so you pay for what your business actually uses.
- Can I define my own action?
- Yes. Give it a name, describe when the assistant should use it in whatever language you like, and declare its parameters. It runs over the gateway you already configured, and you can test it from the panel before going live.
- Can I see what the assistant did?
- Yes. The tool-call log lists every call to your backend with the request sent, the response received, the status and how long it took, so nothing the assistant does is invisible.