Product guide

See how HumanEcho works before you start

Follow the actual product path: create an agent, connect files, test real questions, and embed the assistant on an approved website.

Compare plans View embed example

Quick start

Your first working support pilot

1
Create an agent
Give the support scope a clear name so it is easy to manage later.
2
Attach trusted knowledge
Upload the policies, FAQs, and guides that should ground the answers.
3
Test real customer questions
Use the playground in Arabic and English before publishing the widget.
4
Approve the website origin
Configure the domains allowed to load the public assistant.

Available today

The product surface you can evaluate

This guide describes features present in the current application rather than a future roadmap.

File-based knowledge

Attach supported files and use them as the context for your agent.

Playground testing

Ask production-style questions and review responses before deployment.

Website widget

Add the public assistant to an approved site with a small JavaScript snippet.

Usage visibility

Review message and upload consumption against the active plan.

Website integration

A small embed snippet with explicit controls

Replace PROJECT_ID with the agent you created, choose the locale and appearance, and place the snippet on an approved website.

Allowed-origin rules are configured per project and can also be constrained globally. Test the widget on the exact production origin before launch.

Website embed

<script>
  (() => {
    const script = document.createElement("script");
    script.src = "https://www.humanecho.ai/widget.js";
    script.async = true;
    script.onload = () => {
      window.HumanechoWidget?.init({
        host: "https://www.humanecho.ai",
        projectId: "PROJECT_ID",
        locale: "en",
        theme: { position: "right", accent: "#2563eb" }
      });
    };
    document.head.appendChild(script);
  })();
</script>

Deployment checklist

Verify the controls before customer traffic arrives

A trustworthy launch starts with a narrow knowledge set and an origin policy you have tested.

Keep secrets server-side

Do not place provider keys or backend tokens in the website snippet.

Restrict website origins

Allow only the sites that should be able to load and call the assistant.

Review the source material

Test incomplete, outdated, and ambiguous questions before expanding the pilot.

Build one focused assistant and test it thoroughly

Choose a plan that fits the first support scope, then validate the agent with your own documents and customer questions.

Choose a plan Discuss a team pilot