
Pick the assistant stack by job, not by chatbot marketing:
| Tool | Best job |
|---|---|
| Chatbase | Docs/FAQ-trained AI agent (message credits) |
| Tidio | Live chat + ticketing inbox, with optional Lyro AI conversations |
Use with support workflow tutorial and transcript QA checklist. Price orientation: Cost Calculator. Verify live pricing. Earnings Disclaimer.
Vendor “resolution rate” claims are marketing—treat as unverified for your corpus until you measure reopen/escalation yourself.
| If you need… | Prefer | Why |
|---|---|---|
| A website agent answering from uploaded docs/FAQ | Chatbase | Built around training content + message credits |
| Humans in a shared inbox with optional AI deflection | Tidio | Live chat/ticketing first; Lyro is additive |
| Native agent handoff without a second helpdesk | Tidio | Inbox is core |
| Multi-model / heavier customization experiments | Chatbase | Agent-centric plans (verify model access per tier) |
| Strict weekly QA on transcripts | Either | Process matters more than brand—use the checklist |
| Factor | Chatbase | Tidio |
|---|---|---|
| Core product | AI agents on your sources | Live chat / tickets (+ Flows, Lyro) |
| Free | 50 message credits/mo; inactive agents may be deleted | Free tier with limited conversations; trial available |
| Entry paid (approx.) | Hobby $32/mo annual | Starter ~$24/mo annual |
| AI meter | Message credits (model choice can burn faster) | Lyro AI conversations (separate from human billable chats) |
| Human agents | Seat limits by plan | Up to 10 seats on many self-serve plans |
| Handoff | Often via integrations | Native inbox |
| Risk | Credit surprise + add-ons | Stacking Starter + Lyro + volume tiers |
Chatbase: Upload policy/docs → constrain answers to sources → log failures into a weekly review sheet.
Tidio: Possible with Lyro, but you are paying into a live-chat platform you may not need yet.
Checkpoint: Force “I don’t know / escalate” behavior on missing docs—never invent refunds.
Tidio: Inbox for humans; add Lyro conversation quota for deflection.
Chatbase: Agent for FAQs + separate inbox tool = two systems to QA.
Regardless of vendor, run transcript quality checklist: policy accuracy, citations, handoff notes, reopen rate.
| Pattern | Rough monthly cash | Watch-outs |
|---|---|---|
| Chatbase Free pilot | $0 | 50 credits; agent inactivity rules |
| Chatbase Hobby | ~$32 annual | 500 credits; model mix burns credits |
| Tidio Starter (humans only) | ~$24 annual | Little/no ongoing Lyro without add-on |
| Tidio + Lyro starter AI | often ~$50–90+ | Two meters: human billable + Lyro |
| Scaling either | Variable | Chatbase add-ons; Tidio conversation tiers |
Model stacks in the Cost Calculator and read Free vs Paid.
Chatbase if the product is a docs agent. Tidio if the product is a human inbox that sometimes uses AI. Measure handoffs and wrong answers—do not buy on “automatic resolution” slogans alone.