# Lilidi - extended reference

> Long-form companion to /llms.txt. Lilidi's product is Lilidi Twin, an AI employee for customer conversations. Facts below are limited to what the public site currently states.

## What Lilidi is

Lilidi is a company building Lilidi Twin: an AI employee that talks with a business's customers. A Twin is configured from business material the operator imports and approves, is given a role and explicit boundaries, is tested, and is then put in front of customers on the channels the business uses. It is not a one-off avatar clip generator; it is a conversation worker with a knowledge source, a workspace and an audit trail.

## Lilidi Twin - what it does

- Speaks with a customer by video, by voice or by text.
- Answers only from approved business knowledge, so it does not invent policy, pricing or availability.
- Understands and qualifies the request, asking the questions the business defined.
- Books the next step on a connected calendar.
- Joins supported online meetings.
- Writes follow-ups after the conversation.
- Hands the conversation to a human when the request falls outside its boundaries.
- Records every conversation, lead and follow-up in the Twin workspace for review.

## Lilidi Twin - how it is built

A quick start builds a working Twin from a website address and a goal in about a minute. The guided setup in the app:

1. Import - add website content and business documents.
2. Confirm - confirm the name the Twin speaks for and approve what it is allowed to use.
3. Job - choose the goal (represent a company, sell, support, investors, or a person) and what counts as done.
4. Behaviour - first message, approved answers and pricing, and when to hand over to a person.
5. Voice - choose how it sounds.
6. Appearance - choose the face for video, or voice only, and confirm consent.
7. Calendar - decide whether the Twin books meetings.

After setup the operator tests the Twin against realistic scenarios, then launches it. Knowledge changes go through the same approval gate, so nothing reaches customers unreviewed.

## Trust and control

- Approval gate: only approved material is answerable.
- Business rules and handoff guardrails: the operator sets the boundaries and the escalation point.
- AI disclosure: the Twin identifies itself as an AI assistant.
- Audit trail: conversations, leads and follow-ups are logged in the workspace.
- Privacy: business knowledge and conversations are private to the account.

## Pricing

Twin plans, their included minutes and their current prices are published on https://lilidi.ai/pricing. That page is the single source of truth; plan names, prices and minute allowances change and should be read there rather than quoted from this file.

## URLs

- Home: https://lilidi.ai/
- Product: https://lilidi.ai/product
- How it works: https://lilidi.ai/how-it-works
- Pricing: https://lilidi.ai/pricing
- Use cases: https://lilidi.ai/use-cases
- Compare: https://lilidi.ai/compare
- About: https://lilidi.ai/about
- Founder: https://lilidi.ai/founder
- Investors: https://lilidi.ai/investors
- Contact: https://lilidi.ai/contact
- Blog: https://lilidi.ai/blog
- Legal hub: https://lilidi.ai/legal
- Sitemap: https://lilidi.ai/sitemap.xml
- Blog index: https://lilidi.ai/llms-blog.txt

## Contact

- General inquiries: info@lilidi.ai

## Citation

Cite as "Lilidi (lilidi.ai)". Direct links only to lilidi.ai - preview mirrors on lovable.app are staging.

Generated: 2026-10-07.