Tidio Lyro AI Agent

Lyro is Tidio's AI-powered support agent. Unlike Flows, which follows scripted decision trees, Lyro reads your knowledge base and generates natural-language answers to customer questions on the fly. It is powered by Anthropic's Claude and launched in mid-2023, making it one of the earlier AI chat agents in the SMB customer support space.

The pitch is straightforward: point Lyro at your help center content, toggle it on, and it starts answering questions 24/7 without any manual scripting. For small teams drowning in repetitive "where is my order" and "what is your return policy" tickets, the appeal is obvious. But Lyro is not magic, and there are important nuances around how it works, what it costs, and where it falls short.

Setup process

Getting Lyro running is genuinely fast. The setup involves three steps:

1. Add knowledge sources -- point Lyro at your FAQ pages, help center URLs, or manually enter question-answer pairs. Lyro crawls the content and builds its understanding from there.
2. Configure behavior -- set the tone, define when Lyro should hand off to a human, and choose which channels it operates on.
3. Activate -- toggle Lyro on for your chat widget.

Most reviewers on G2 report having Lyro answering questions within 30 minutes of starting setup. There is no model training involved on your end -- Lyro uses retrieval-augmented generation, pulling relevant content from your knowledge base at query time and generating a response. If your help content is well-organized, setup is nearly zero-effort.

The catch is that Lyro is only as good as the content you feed it. If your FAQ pages are outdated, vague, or incomplete, Lyro's answers will reflect that. Several reviewers note that they spent more time cleaning up their help content than actually configuring Lyro itself -- which is arguably time well spent regardless.

Capabilities

Lyro has expanded significantly since its initial launch. Here is what it can do as of early 2026:

Core conversation handling:

• Answers customer questions using your knowledge base content
• Supports follow-up questions within a conversation thread
• Provides source links so customers can read the full article
• Responds in under 6 seconds on average
• Supports 12 languages (auto-detects and responds in the customer's language)

Smart Actions:

• Lyro can perform tasks beyond just answering questions. Smart Actions let it connect to external data -- the most notable is Shopify order lookup, where Lyro can check order status, tracking information, and delivery estimates directly in conversation.
• Product recommendations based on browsing history and catalog data
• Discount code delivery based on conversation context

Channel coverage:

• Web chat widget
• Email
• Facebook Messenger
• Instagram DMs
• WhatsApp

Conversation management:

• Automatic human handoff when Lyro is not confident in its answer or when the customer requests a human
• Configurable confidence thresholds for escalation
• Conversation summaries passed to agents on handoff so they have context

How Lyro learns

Lyro does not learn from individual conversations in real time. It learns from the knowledge sources you provide:

• URLs -- Lyro crawls web pages (help center articles, FAQ pages, landing pages) and extracts information
• Manual Q&A pairs -- you can add specific question-answer pairs for topics not covered by existing content
• Document uploads -- PDF and text file uploads for internal documentation

When a customer asks a question, Lyro searches your knowledge base for relevant content, retrieves the most applicable information, and generates a conversational answer. It does not memorize customer interactions or update its knowledge based on what agents tell customers. If you want Lyro to know something new, you add it to the knowledge base.

This is both a strength and a limitation. The strength is predictability -- Lyro will not suddenly start saying things you did not intend because of a weird training loop. The limitation is that keeping Lyro current requires actively maintaining your knowledge sources. If your shipping policy changes and you do not update the FAQ page, Lyro will keep giving the old answer.

Resolution rates and performance

Tidio reports average resolution rates of 64-67% across its customer base, with some businesses reporting up to 90%. These numbers are in line with what I see in user reviews -- businesses with clean, comprehensive knowledge bases consistently report higher rates.

What counts as a "resolution" matters here. Lyro marks a conversation as resolved when the customer confirms their question was answered or when the conversation ends without escalation. It is a reasonable metric but not a perfect one -- a customer who gives up and leaves is not the same as a customer who got a helpful answer. Tidio does provide CSAT tracking for Lyro conversations on higher plans, which gives a more honest picture.

Response speed is a consistent bright spot. Multiple G2 and Capterra reviewers highlight that Lyro responds significantly faster than human agents, and that customers rarely complain about the quality of AI-generated answers for straightforward questions.

Pricing

This is where Lyro gets complicated, and it is the most common source of confusion in reviews.

The 50 free Lyro conversations are a one-time allowance, not a monthly refresh. Once you use them, they are gone. After that, you either subscribe to a Lyro add-on or upgrade to Plus.

Lyro add-on pricing:

At lower volumes, Lyro costs roughly $0.70-$0.78 per AI conversation. This is cheaper than Intercom's Fin at $0.99 per resolution, but more expensive than handling the conversation with a human agent on a per-seat plan if your agents are not at capacity. The economics make the most sense for businesses where Lyro is genuinely deflecting conversations that would otherwise require agent time.

The Plus plan ($749/mo and up) includes Lyro with no per-conversation cap, which makes it the better option for businesses needing heavy AI usage -- but the jump in base price is significant.

What reviews praise

Across G2, Capterra, and Trustpilot, Lyro receives generally positive sentiment, particularly from small and mid-size ecommerce businesses:

• Fast setup for non-technical teams -- the most consistent praise. Business owners without any technical background report getting Lyro running in under an hour.
• Good deflection of routine questions -- shipping, returns, account issues, product specs. The kinds of questions that make up 60-70% of support volume for most ecommerce stores.
• Multi-language support -- businesses serving international customers appreciate that Lyro auto-detects language without needing separate configurations.
• Human handoff works well -- when Lyro cannot answer, the transition to a live agent is smooth and includes conversation context.
• Shopify integration depth -- the ability to look up orders and recommend products within conversation is cited as a significant differentiator.

What reviews criticize

The criticism is also fairly consistent:

• Hallucination concerns -- several reviewers mention instances where Lyro generated answers that were plausible-sounding but inaccurate. This is an inherent risk with generative AI, and Tidio has added guardrails (confidence thresholds, source linking), but it has not been eliminated. Businesses in regulated industries or with complex product lines need to monitor Lyro's output carefully.
• Limited customization -- you can set the tone and add knowledge, but you cannot deeply customize how Lyro structures responses, what it emphasizes, or how it handles edge cases. Some reviewers want more control over the AI's personality and decision-making.
• Cost adds up at scale -- the per-conversation pricing on add-on plans means costs grow linearly with volume. A business handling 1,000 Lyro conversations per month would pay roughly $500+ on add-on pricing, at which point the Plus plan starts to make more sense.
• Knowledge base maintenance burden -- Lyro's answer quality is directly tied to content quality. Businesses that do not actively maintain their help content see declining performance over time.
• One-time free allocation is misleading -- multiple reviewers express frustration that the "50 free Lyro conversations" are not monthly. The marketing does not always make this clear, and users feel surprised when the allocation runs out.

Comparison to dedicated AI tools

Lyro competes with a growing field of AI support agents. Compared to Intercom's Fin, Lyro is cheaper per conversation and easier to set up, but Fin supports more channels (including voice), offers more advanced multi-step procedures, and has stronger enterprise compliance credentials. Compared to standalone tools like Ada or Forethought, Lyro is less configurable but comes bundled with Tidio's broader platform -- live chat, Flows, ticketing -- which means fewer tools to manage.

For businesses already using Tidio for live chat, adding Lyro is a natural extension that avoids the integration complexity of a standalone AI tool. For businesses evaluating Tidio primarily for Lyro, it is worth comparing the total cost (Tidio plan + Lyro add-on) against alternatives that might include the AI agent in their base pricing.

Bottom line

Lyro is a capable AI support agent that delivers real value for small to mid-size businesses, particularly in ecommerce. The setup is genuinely fast, the Shopify integration is a standout, and for routine questions it performs well. The main concerns are around cost transparency (that one-time free tier catches people off guard), hallucination risk that requires ongoing monitoring, and the limited customization options for businesses with specific needs.

For most Tidio users, the best results come from pairing Lyro with Flows -- let Flows handle the structured, predictable interactions (cart recovery, lead qualification, routing) and let Lyro handle the open-ended questions. That combination covers more ground than either tool alone.

Last updated: April 2026.