Chatty AI Chatbot & Live Chat — Internal Review Research & Evidence

INTERNAL WORKING FILE. This is not the publishable review. It records the supplied hands-on evidence, public research, claim boundaries, unresolved questions, and publication gates behind reviews/chatty.md.

Review Record

• App: Chatty AI Chatbot & Live Chat (install dialog: “Chatty — Chatty AI Sales Agent - For live chat & helpdesk”)
• Developer (listing): Chatty AI Sales Agent - For live chat & helpdesk; address Peace Centre, Singapore
• Category: Customer Support and Helpdesk — fifth review in this category (after Chatra, Tidio, Zipchat, Willdesk)
• Human reviewer / assignee: Munna
• ClickUp task: Not provided in the session
• Shopify App Store: https://apps.shopify.com/chatty
• Research date / last verified: 2026-10-12
• Content Preview URL: https://reviews.blinto.co/reviews/chatty/
• Test store: zenden (zenden-6975.myshopify.com), Horizon theme, 9-product jersey catalogue (all tagged best_selling and newest), ~1,700 orders (admin 1,707). Same store as the Willdesk review.
• Requested constraints: quality score above 95; read time not more than 14 minutes. Delivered at 2,519 body words (13 min at 200 wpm).

Human Testing & Evidence

Testing level

• Level 1 — Full Hands-On Test (install → onboarding → Free plan → embed → AI training → storefront questions → human handover → real order tracking → inbox)

Merchant decision question: Does Chatty’s AI agent answer real shopper questions correctly on the Free plan, and does the hand-off to a human work when it cannot?

What was tested (supplied captures, in order)

1. Storefront chat: product card Japan Away Kit Player Full Sleeve $1,299, Add to cart; shopper “What is the most selling product”; AI: “The Japan Away Kit Player Full Sleeve is tagged as a best-selling clothing product. It’s a unisex, long-sleeve football jersey made from breathable polyester, available in M, L, XL, and 2XL for USD 1,299. I can’t verify whether it’s the single top-selling product in the store.”
2. Inbox: header “Customer • Customer for 2 months • 0 orders • $NaN spent amount”; shopper messages tagged Product recommendation and Product price; Feedback / Review sources under AI replies; “AI is processing, joining conversation will stop AI mode” + Join conversation; left nav Conversations (All, Your inbox, Unassigned, Starred, Blocked), Chatty AI (Processing 1, Handover, Resolved), Channels (Online store, Email, Messenger, Instagram, WhatsApp); details: Assignee AI Agent; Visitor device Edge 155 on Windows; Recent orders #2757 10/12/2026 Paid Unfulfilled $1286.01, #2756 10/12/2026 $2598, #2755 10/09/2026 $1294.
3. Inbox (continued): AI: “I can’t verify which product is the cheapest in the store from the information I have. I don’t want to guess based on the one clothing item we discussed earlier. Are you looking for the cheapest clothing item, or the cheapest item across the whole store?”; shopper “cheapest clothing item”; AI: “I can’t verify the cheapest clothing item from the product information available to me. The only clothing item we’ve discussed is the Japan Away Kit Player Full Sleeve, listed at USD 1,299, but there may be less expensive items. I don’t want to mistake it for the cheapest.”; system: “This conversation was reassigned to Agent at 10/12/2026 12:59 PM”; team member reply: “Cheapest Price is: 1. Norway World Cup 2026 Home Kit Player half $999.00 2. Jamaica Away Player $999.00”; Assignee now the team member (Inactive).
4. Storefront: the same team-member reply in the widget, with the store’s “ZE” avatar.
5. Install dialog: view customer data (sensitive, device and activity), staff and contributor data; view and edit customers, products, orders, discounts, store analytics, Online Store, Shopify admin, other data.
6. Onboarding 1: “Which of these best describes you?” — AI-powered (selected; “AI handles all conversations 24/7. Your team steps in only when needed.”), Human only, Hybrid; language English.
7. Onboarding 2 — Your brand, your tone: 11 brand colours (purple selected), AI name “Chatty AI”, welcome “Hello {{customer_name}} 👋 I’m Chatty AI agent, here to help you find what you’re looking for. How can I help you?”, language English; live preview.
8. Onboarding 3 — How AI transfer to your team: Always transfer to team (selected), Only when team is online, Never transfer; preview of “Can I talk to human?” handover.
9. Onboarding 4 — Try these use cases: Product recommendation (selected), Order tracking, Returns & Refund; preview; Finish.
10. Plan screen: Monthly/Yearly (save up to 15%). Basic $19.99/mo, 7-day trial, 100 AI conversations/month, AI order tracking, multi-channel inbox, 12 months history, basic analytics. Pro $48.29 (30% OFF LIFETIME, from $68.99), Recommended, +2,000 lifetime AI conversations ($800 value), 500/month, AI shopping assistant, proactive engage, auto-recover abandoned carts, AI replies on social, advanced analytics, unlimited history. Plus $139.30 (from $199), +2,000 lifetime, 1,000/month, unlimited product training, multi-store, dedicated AI consultant, priority support, “Contact us for better deals”. Free: 50 AI conversations/month, unlimited human conversations, AI trained on up to 200 products, AI answers FAQs & recommends products, email connect, widget customization, after-hours auto-reply + saved replies; ✗ multiple agents, ✗ fully-customizable chatbox. Continue as Free selected.
11. Theme embed: “Chatty - Core”, no customizable settings; launcher bubble.
12. Training data — Products: 9 of 9 products learned; auto sync daily; last updated 10/12/26 12:37 PM; tags best_selling + newest on every product; FAQs 0 each; “Upgrade to bulk edit status”.
13. Training data — Knowledge base: 10 entries, 2 issues, FAQ sync finished; 10 Live · 0 Off; FAQs 9 synced; Store pages 1 page synced · 3 pages on store · 2 of 10 syncs today; AI suggestions 0, Answer gaps 0, Review queue 0; entries incl. “Your Privacy Choices”, “Do you accept store credit?”.
14. AI agent page: “AI agent On”; “You are almost ready — ✓ Training data, Instructions, ✓ Skills, Scenarios, ✓ Test AI”; Training data 10 knowledge entries; Instructions; Skills 4 of 10 live; Scenarios; AI unresolved questions; Settings. App nav: Inbox, Contacts, FAQs, Chatbox, AI agent, Proactive chat, View more.
15. Storefront widget: welcome message; quick replies Track my order, How long does delivery take?, Return policy.
16. Gmail: “Order #2757 confirmed” — Brazil World Cup 2026 Away Kit Player Edition Full, placed Mon, Oct 12.
17. Storefront tracking: greeting by name; Track my order → “You can track your order using the tracking number provided in your shipping confirmation email. Click on Track navigation”; shopper “2757” → “Order #2757 is paid and still processing; it hasn’t shipped yet. You can check for updates on your order status page.”
18. Training data — Products (full list): Argentina, Brazil, England, Jamaica, Japan, Norway, Portugal Home, Portugal Away, Spain — all Active, all tagged best_selling/newest; “Settings saved”.

What was not tested

• WhatsApp, Messenger, Instagram, email channels; proactive chat; FAQ help center; checkout chat
• Custom Instructions and Scenarios (both open on the setup banner); Skills beyond the 4 live
• Returns & Refund flow; delivery-time quick reply
• Analytics, mobile app, multi-agent features, any paid plan, trial, overage or billing
• What resolves the 2 knowledge-base “issues”

Testing limitations

• Unknown whether the reassignment to an agent was triggered automatically by Chatty (Always transfer to team) or by the team member joining from the inbox. The article says only that the conversation “was reassigned to an agent”. Ask Munna.
• No custom instructions were written before the catalogue question; the article says this may change the result.
• The team member who answered is a Blinto colleague; the answer was checked against the storefront grid ($999 for Norway and Jamaica; $1,299 for the others visible).
• AI order tracking worked on the Free plan although the in-app plan table lists it as a Basic addition; reported as observed, not explained.

Evidence Inventory & Public Authenticity Plan

Public evidence safety

• Customer name, email and avatars pixelated in both inbox captures.
• Team member’s name and avatar pixelated.
• Shopper’s name in the widget greeting and the storefront account initial pixelated.
• Chatty’s floating support-agent photo pixelated on admin captures.
• Full-frame, downscaled to 1,400 px; Windows activation watermark left in place.
• Editorial thumbnail generated and saved. The sky emblem rendered as a soft lavender cloud rather than a distinct speech-bubble shape; alt text describes the image as delivered.

Claim → Evidence Check

Official Sources

• Shopify App Store listing: https://apps.shopify.com/chatty — checked 2026-10-12. Free: 50 AI conversations/month, unlimited human conversation, AI trained up to 200 products, AI answers FAQs & recommends products, email connect, widget customization, after-hours auto reply + saved replies. Basic $19.99/mo ($203.88/yr, save 15%), $0.4 per additional AI conversation, 100 AI conversations, 12-month history, AI order tracking, 5 members, multi-channel inbox, translation, standard AI model, basic analytics, 7-day trial. Pro $68.99 ($707.88/yr), $0.4 overage, 500 AI conversations, unlimited history, AI shopping assistant, Pro AI model, proactive engage, abandoned carts, advanced analytics, AI reply on socials, 7-day trial. Plus $199 ($2,039.88/yr), $0.4 overage, 1,000 AI conversations, unlimited members, train AI with up to 20,000 products, Pro AI model, AI shopping assistant, AI analytics consultant, 7-day trial. “Recurring and usage-based charges are billed every 30 days.” Rating 4.9 (1,877): 96% five, 57 four, 5 three, 6 two, 16 one. Launched June 9, 2022. Built for Shopify. Works with Checkout, Shopify Flow, Shopify Admin, Email, Instagram, Klaviyo, Messenger, WhatsApp, Zendesk. Languages: English, German, French, Spanish, Japanese, Portuguese (Portugal), Chinese (Simplified).
• Reviews (raw HTML: 1–3-star, 4-star, newest): https://apps.shopify.com/chatty/reviews — checked 2026-10-12.
• Prior Blinto reviews used for comparisons: reviews/tidio.md, reviews/zipchat.md, reviews/willdesk.md, reviews/chatra.md.

Merchant Feedback Patterns

Positives (Sep–Oct 2026): fast, friendly support from named staff (Hazel, Jade, Liz, Andy, Mirra); setup help; fixes such as a chat window disappearing when customers tried to start a conversation; customisation; a useful free plan.

Criticism (1–3-star feed, all with developer replies):

• KICKSOWN (US, 8 months), Oct 7 2026: “an extra $68 every month on top of my subscription fee”. Reply: Pro includes 500 AI conversations per cycle; above that the AI pauses or continues at $0.40 each up to a configurable monthly overage limit.
• HerScore Solutions (US), Jul 24 2026: charged full price instead of a trial; reply: store had activated the app before.
• ElGranero Typewriter.Company (Spain, ~2 years), Jul 10 2026: quality falling as prices rise; loops; mobile logouts; FAQ not recognised by Google Search Console. Reply: logout fix in testing; FAQ issue due to disabled embed.
• Chic Marine (Canada, ~1 year), Jun 12 2026: lost a day of customer support. Reply: no system error found; suspected browser extension.
• AgozTech (US, 5 months), Feb 10 2026: AI rewrote a “Start a Return” link.
• eVolt.si (Slovenia, 6 days), Nov 26 2025: site slowdown; predefined answers not clickable; chats failing to start. Reply: two bugs fixed.
• Nyche Hair (US, 4 months), Nov 10 2025: blank screen / “website blocked”; AI unhelpful until the owner joined the chat.
• Majos (India, 8 days), Nov 1 2025: historical chats; “Hi Null”. Reply: fixed.

Interpretation boundary: ~1.4% of ratings at three stars or below; none reproduced. Our AI was cautious rather than wrong.

Missing Evidence / Questions

• Was the reassignment to an agent automatic, or did the team member join/reassign manually?
• What are the 2 knowledge-base “issues”?
• Would custom Instructions let the AI answer the cheapest-product question?
• Does AI order tracking stay available on Free?
• Munna’s subjective impressions.
• ClickUp task URL.

Editorial Evaluation

Strengths: guided setup with live preview; honest AI caveats; correct order tracking; working handover; organised inbox with intent tags and recent orders; generous Free plan.

Limitations: declined a simple price comparison; NaN/0-orders header bug; $0.40 overage; listing vs in-app price differences; order-tracking plan discrepancy; sourced reliability complaints.

Best fit: Small and growing stores wanting AI plus a human inbox, preferring hand-offs to guesses, mostly order/policy questions, willing to write instructions.

Poor fit: Stores needing the AI to answer catalogue price/comparison questions unaided; teams that will not monitor AI overage.

Claims We Must Not Make

• Do not say Chatty transferred the chat automatically; the trigger is unrecorded.
• Do not say the AI cannot answer price questions in general; one session, no instructions.
• Do not claim channels, proactive chat, FAQ center, returns, analytics or paid features work.
• Do not present sourced complaints as reproduced.
• Do not rank Chatty against Tidio/Zipchat/Willdesk on the basis of single questions.
• Do not state the 30% lifetime discount applies to all merchants.

Blinto Review Quality Score — Internal

Status: 90–100 — Strong, publishable. Scored 2026-10-12. Clears the 81+ auto-approval threshold in AGENTS.md §10. Provisional until Munna’s fact-check is complete.

Editorial Verdict Lock

• Verdict locked: Pending Munna’s fact-check.
• Decision: Recommended for small and growing stores wanting an honest AI agent with a real inbox, provided instructions are written and an overage limit is set.

Founder Intelligence — Internal

• Exceptional: setup with live preview; AI caveats instead of hallucination; order-status answers; inbox context (intent tags, recent orders, source review).
• Friction: AI declines simple catalogue comparisons it has the data for; NaN/0-orders header; overage surprises; price presentation differs between listing and app.
• Opportunities: allow structured catalogue queries (min/max price) from synced product data; fix the customer-summary header; surface overage usage and the cap prominently; align listing and in-app pricing.
• Questions for the founder: Why can the agent not compare prices across learned products? What do the knowledge-base “issues” flag? Is AI order tracking intended on Free?
• Blinto fit: AI capability + billing-clarity opportunity.

SEO & Content Direction

• Primary keyword: Chatty review. Secondary: Chatty AI chatbot Shopify, Chatty live chat pricing, Chatty vs Tidio, Chatty vs Zipchat, Shopify AI chatbot.
• Intent: Commercial investigation.
• Angle: The category’s standing cheapest-product test, with Chatty’s “won’t guess” outcome and a working human handover.
• Internal links: /reviews/tidio/, /reviews/zipchat/, /reviews/willdesk/, /reviews/chatra/.
• Disclosures: Level 1; development store; Free plan only; channels, proactive chat, instructions and paid features untested; data checked 2026-10-12.

Thumbnail Concept and Generation Prompt

Concept: A shopkeeper behind a counter pauses, hand raised to an approaching assistant, rather than pointing a customer somewhere at random. An open ledger and a row of identical jars sit on the counter. Reads as “it won’t guess; it asks for help.”

Prompt:

Generate a single image and export it as a PNG file, 16:9 landscape, 1792×1024 pixels or larger.

This must look like a soft, hazy, atmospheric vintage oil painting — NOT a sharp digital illustration. Softness and abstraction are the most important requirement.

Style: a sun-bleached impressionist oil painting on rough canvas, in the manner of a faded plate from an old art book. Thick, loose, visible impasto brushstrokes everywhere, including the foreground. Strong canvas texture and pigment variation across the whole surface. Low contrast, high-key light: warm cream, ivory, pale ochre, dusty gold, muted olive, soft grey-blue. No pure blacks, no deep shadows, no saturated colour. Everything slightly out of focus, as if seen through warm afternoon haze; edges lost and found, silhouettes soft, forms suggested rather than drawn.

Scene: inside a sunlit open-fronted village shop, a small shopkeeper figure stands behind a wooden counter lined with a row of identical clay jars. A customer waits at the counter. The shopkeeper turns and raises one hand to beckon a second figure walking in from a doorway behind, as if calling a colleague over to answer. The figures are small in the frame and have no facial detail; the shop interior dissolves into broad blurred masses of colour.

Sky: through the wide open shopfront, in the upper background, a large soft rounded speech-bubble shape formed from haze and cloud glows in pale pastel lavender-violet, around a fifth of the canvas width — feathered edges, varied opacity, painted with the same loose brushwork, discovered in the clouds rather than placed on top, never a logo or a clean outline.

Depth: foreground moderately soft, middle ground much softer, background almost abstract colour fields.

Strict rules: no text, no letters, no numbers, no logos, no signage, no labels on jars, no UI elements, no screens, no phones, no watermark. No photorealism, no 3D render, no glossy digital painting, no fine detail, no sharp edges anywhere, no faces with detail. Wide composition with generous empty space left and right so the image crops safely.

Output format: PNG.

Source colour note: Chatty’s icon is a rounded chat bubble in deep blue-violet, and its default widget header is purple; the emblem is the bubble shape softened to pastel lavender-violet. Distinct from the support set’s existing emblems (verify against Tidio, Zipchat, Chatra and Willdesk thumbnails before generating).

Public Article Authenticity Gate

• What We Tested visible and accurate.
• Eight purposeful screenshots embedded; files exist.
• Findings map to captures in this record.
• Developer claims, merchant feedback and Blinto observations distinguished.
• Cross-app comparisons bounded as single questions / published limits.
• Evidence redacted and publication-safe.
• Read time within 14 minutes (13 min).
• Editorial thumbnail generated and added.
• Rendered on the Review Hub; Munna fact-checked.

Publication Gate

• Hands-on claims supported by captures.
• Listing facts verified 2026-10-12.
• Handover trigger qualified rather than asserted.
• Quality Score 97/100 (provisional) — clears auto-approval.
• Content Preview URL added to ClickUp.

The internal research preview is /research/chatty/. Any SEO handoff must use /reviews/chatty/.