From URL to live agent

Quickstart

Five steps, about five minutes. Nothing here needs a developer until the widget install — and that’s one line

1. 01

   Paste your URL

   Nerv reads your site, indexes the pages and builds the first knowledge base automatically. The onboarding chat asks a few questions — business type, channels, tone — and assembles the rest

2. 02

   Review the knowledge base

   Check what was indexed, drop in extra docs if you have them, and remove anything the AI shouldn't answer from. Knowledge gap suggestions will flag missing topics as real questions come in

3. 03

   Install the widget

   One script tag on your site. Theme, colour and position are configured from the workspace — no redeploy needed after the first install

4. 04

   Connect channels

   Email, Telegram, WhatsApp and Meta connect from workspace settings. Every channel lands in the same inbox and the same AI pipeline — on every plan

5. 05

   Invite your team

   Add agents to your included seats. Nerv hands conversations to them with a structured brief — context, customer history and what was already tried

Install the widget

Drop the snippet before the closing </body> tag. Your App ID is in Settings → Widget. Keep type="module" — the SDK ships as an ES module and finds its own install tag by data-app-id.

<script
  type="module"
  src="https://cdn.chatly.tech/widget.js"
  data-app-id="YOUR_APP_ID">
</script>

Theme, accent colour, position, greeting and proactive triggers are all configured from the workspace and update live — install once, never touch the code again.

Configure Nerv

Everything about how Nerv looks and behaves is set from the workspace and updates live — no redeploy

Theme

System, Light or Dark — System follows the visitor's OS

Brand colour

Any hex value for the launcher, header and customer bubbles

Launcher

Circle, Rounded or Square, in any of the four corners

Header style

Minimal, or Full with an avatar and online status

Help center

Show or hide the searchable Help tab and its collections

Welcome & starters

Greeting, home subtitle, search and conversation starters

Proactive triggers

Open the messenger with a targeted message on a given page

Branding

The “Powered by TruSend” badge — removable on Pro

How Nerv answers

Nerv answers from your help center and past conversations using hybrid retrieval (BM25 + vector) with reranking — so replies are built from your real content, not a generic model’s guess.

Every answer carries a confidence signal. When Nerv isn’t sure — or a customer asks for a person — it hands the conversation to your team with the full thread instead of guessing. In the workspace you can open any reply and see the sources and decisions behind it.

Channels & hand-off

Connect Email, Telegram, WhatsApp and Meta from workspace settings. Every channel lands in one inbox and runs the same AI pipeline — on every plan.

When Nerv hands a conversation to a human, your team gets a structured brief — context, history and what was already tried — not a wall of messages to re-read. Hand-offs to your team are never billed.

Billing & limits

Plans include a monthly allowance of resolves — conversations Nerv closes end-to-end. Hand-offs to your team are never billed. Past the allowance, overage bills at the rate printed on your plan; an email alert fires at 90% and an optional hard cap stops the AI instead of the spend.

Fair-use on AI conversations is ratio-based: only conversations beyond ≈3.3× the resolves the AI actually delivers bill $0.02 each, so at any healthy AI Handle Rate it’s $0 — at any volume. The guides published per plan (500 on Starter, 2,667 on Growth, 10,000 on Pro) are that formula at each plan’s included resolves. You only pay it when the AI is attempting far more conversations than it resolves — a low AI Handle Rate — which is the signal to feed the knowledge base, not necessarily to upgrade.

Rates, seat counts and a cost calculator live on the pricing page.

API & MCP

Coming soon

REST API access is included on Growth and above — keys are issued per workspace in Settings → API. The full endpoint reference, and our MCP server, are being written alongside the product itself. If you need something specific now, ask at support@trusend.ai and you’ll get a real answer from the team that builds it.