Zipchat 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/zipchat.md.

Review Record

• App: Zipchat AI Chatbot & Live Chat (Shopify install dialog and admin show “ZipChat: AI Chatbot”; theme embed “Zipchat Bubble”)
• Developer: ZIPCHAT INC
• Category: Customer Support and Helpdesk — second review in this category, after Tidio
• Human reviewer / assignee: Munna
• Due date: Not provided
• ClickUp task: Not provided — Batch 3, SOP-1 score 42/50, Priority P1
• Shopify App Store: https://apps.shopify.com/zipchat
• Research date: 2026-10-07
• Last verified: 2026-10-07
• Content Preview URL: /reviews/zipchat/ on reviews.blinto.co
• Published: 2026-10-07 — https://blinto.co/blog/zipchat-shopify-review/ (live read time 13 min; published on the blinto-next front end, which renders Key Takeaways as its own component rather than an H2 — content identical)
• Test date: Conversation record reads 05:45 – 07 October 2026; dashboard performance range Sep 08 – Oct 07, 2026
• Test store: Nestly (nestly.blinto.co), Tinker theme — the same store, catalogue and theme used for the Tidio review, which is what makes the comparison in this article controlled. Disclosed in the article.
• Requested constraints: quality score above 90; read time 12–14 minutes. Delivered at 2,677 body words (13.4 min at 200 wpm; the public site rounds up, as it did for Tidio at 2,733 words → 14 min).

Human Testing & Evidence

Testing level

• Level 1 — Full Hands-On Test (install → onboarding → embed → live shopper conversations → cart → merchant view)
• Level 2 — Partial Hands-On Test
• Level 3 — Demo and Documentation Review
• Level 4 — Research-Only Evaluation
• Level 5 — Insufficient Evidence

Merchant decision question: Can Zipchat’s AI agent answer catalogue questions correctly and convert a shopper question into a cart — and what does that capability cost?

Realistic scenario tested: On the Nestly development store (Tinker theme, 51 low-priced accessories), install Zipchat, accept the default agent behaviour, enable the Zipchat Bubble app embed, then as a storefront shopper ask (a) a deliberately ambiguous value question with no product named and (b) the same comparative catalogue question previously put to Tidio on this store; add the recommended product to the cart from inside the chat; verify the price on the product page and in the cart; then inspect the conversation in the Zipchat admin.

What was tested (from supplied screenshots)

1. Shopify install dialog: “ZipChat: AI Chatbot, ZIPCHAT INC”; permissions — view customer data (sensitive data, device and activity data), view staff and contributor data (store owner, blog contributors), view and edit store data (customers, products, orders, discounts, store analytics, Online Store). Store “Nestly”.
2. Dashboard launch checklist (“Get your agent live on your website”): step 1 “Zipchat learns your store” already complete (green tick); step 2 “Teach your agent how to behave ~1 min” with the note “Your agent is studying nestly.blinto.co right now to match its tone of voice to your brand. Usually about a minute.”; “What should your agent do?” — ☑ Hand off to a human when unsure (Recommended), Notify you by Email / In app only (selected); ☑ Offer a discount to hesitant buyers, Maximum discount 20%; ☑ Collect emails from interested visitors; “Write your own rules (advanced)”; Save behavior. Steps 3 and 4: “Meet your agent ~2 min”, “Go live on your website ~3 min”. Performance panel dated Sep 08 – Oct 07, 2026, All channels, “Revenue: 3 influence levels”; Status — AI replies: Active. Sidebar: Conversations, Contacts, Campaigns, Reports, AI training, Channels, View more. (Captured twice, captures 2 and 3.)
3. Shopify theme editor — App embeds (Nestly, Tinker, Active): search “ZipChat: AI Chatbot” → “Zipchat Bubble — ZipChat: AI Chatbot” toggled on; “This app embed doesn’t have customizable settings. Manage app.”; storefront preview shows a pill-shaped “Ask AI…” bar centred at the bottom with image and microphone icons.
4. Shopify Online Store page: Nestly, nestly.blinto.co, Tinker, “Last saved: Just now”; Web Vitals panel LCP P75 1388 ms (Good), INP P75 40 ms (Good), CLS 0.01, Sessions by Device 113 Desktop / 2 Mobile; ZipChat listed under Apps. Not attributable to Zipchat — store-level metrics only; not used in the article.
5. Storefront home page: the “Ask AI…” bar visible at the bottom centre.
6. Conversation 1 — ambiguity: agent opens “Welcome to Nestly, how can I help you today?”; shopper: “💰 What makes this worth the price?”; agent: “Hello! I’m Nestly, your AI assistant. 😊 Could you please clarify which product you’re asking about?”; shopper: “Which product are you asking about? Acrylic Keyboard Stand”; agent: “The Acrylic Keyboard Stand ($4.87 – $5.34 USD) offers great value for your workspace setup! ✨ Key benefits include: — Durable Material: crafted with quality acrylic…; — Compact Desktop Lifter: designed to elevate your keyboard or mini PC…; — Sleek & Versatile: elegant compact design available in Black and Ivory…”. Footer: “AI agent by Zipchat”.
7. Conversation 2 — the controlled question: shopper: “What is the cheapest product in this store”; agent: “The cheapest product in the store is the 1 Piece Dolls House Accessory Miniature Digital SLR Camera Model, priced at just $1.18 USD! 📷 Another very budget-friendly option is the Cute Simulation 1:12 Mini Laptop Computer Model starting at $1.29 USD.” Two product cards rendered beneath with images, titles, $1.18 and $1.29+, each with an add-to-cart icon.
8. In-chat product quick view: the camera model with image, full title, “Learn more”, variant “Default Title”, $1.18 and an Add to cart button, inside the chat panel; footer “AI agent by Zipchat”.
9. After adding: the $1.18 product card’s cart icon replaced by a tick; header cart badge shows 1.
10. Product page: “1 Piece Dolls House Accessory Home Decoration For Doll Dollhouse Miniature Digital SLR Camera Model”, $1.18 highlighted, Add to cart / Buy it now, SPECIFICATIONS, Brand Name: JULYHOT. Confirms the agent’s answer against the storefront.
11. Cart drawer: the same product, Default Title, $1.18 USD, quantity 1, Estimated total $1.18 USD, Check out.
12. Zipchat Conversations admin: tabs Active / Escalated / Assigned to me / Archived; conversation list entry (customer name — PII) 13 min, preview “What is the cheapest product in this sto…”; Conversation history with both answers; “Debug” and “Make correction” controls under every AI message; Mark unread / Archive / Take over; Assignee “Nestly”; Customer panel with name and email (PII), channel Zipchat Bubble, “05:45 - 07 October 2026”, https://nestly.blinto.co; Insights — Customer sentiment: Pending; Share URL https://app.zipchat.ai/chats/21892/; footer “Zipchat will respond to the next messages automatically”.

What was not tested

• WhatsApp, Instagram, Messenger and email channels
• Campaigns, Reports, Contacts, AI training screens, custom rules (“Write your own rules”)
• Human take-over, escalation, the Debug and Make correction controls themselves
• Multilingual answers; mobile rendering of the Ask AI bar
• Checkout completion (cart reached, order not placed)
• Any listed third-party integration (Gorgias, Zendesk, Re:amaze, Klaviyo, Omnisend, Brevo)
• Order tracking, returns, COD verification, stock checks, cart recovery, upsell campaigns
• Any billing event — trial end, paid plan, overage, downgrade or uninstall
• Zipchat’s support team; the discount behaviour actually firing
• Whether the agent’s answers change after a correction

Testing limitations

• Development store with a 51-item accessories catalogue; no real customers or traffic.
• A handful of questions in a single session. The accuracy result is an observation, not a measured rate, and the article says so twice.
• Default behaviour accepted; no custom rules written, and the 20% discount behaviour was never triggered or observed firing.
• Plan state not captured — no pricing or plan screen was photographed in-app, so which tier the store was on is unknown.
• The Web Vitals figures on the Shopify Online Store page are store-level and predate/overlap other apps; they are not attributable to Zipchat and are excluded from the article.
• The developer’s description of Admin API catalogue ingestion is a vendor claim used to explain, not prove, the observed result.
• Reviewer’s subjective impressions were not supplied; the article describes observable behaviour only.

Raw observations / experience

• “Zipchat learns your store” was already complete on arrival at the dashboard — no manual knowledge step, in contrast to Tidio’s single-URL field.
• Onboarding ships three behaviours enabled: hand-off when unsure, discount to hesitant buyers (max 20%), collect emails.
• The theme embed has no settings; the storefront surface is a centred “Ask AI…” bar, not a corner bubble.
• The agent asked a clarifying question rather than answering an ambiguous prompt.
• The agent answered the comparative catalogue question correctly ($1.18 primary, $1.29 runner-up), verified against the product page and cart.
• Product cards carry add-to-cart; the item entered the real Shopify cart at the correct price without leaving the chat.
• Merchant inbox exposes Debug and Make correction per AI message, plus Take over and a shareable conversation URL.
• Customer sentiment showed as “Pending” rather than a value.

Evidence Inventory & Public Authenticity Plan

Thirteen captures were supplied in chat and recovered from the session transcript on 2026-10-07. Eight are published full-frame (downscaled to 1400px, personal data pixelated); the rest are supplemental internal evidence.

Optional screen recording: Not needed, though a recording of the two conversations would strengthen the comparison if it is ever challenged.

Public evidence safety

• Reviewer name and email pixelated in the Conversations admin capture, in both the conversation list and the customer panel.
• Storefront account initial pixelated on all published storefront captures.
• Store names Nestly and nestly.blinto.co left visible, consistent with prior reviews.
• The Zipchat share URL (app.zipchat.ai/chats/21892/) is visible but contains no personal data and is not reachable without authentication; retained because it evidences the share feature.
• Selected screenshots come from the actual test sequence.
• No selected screenshot is decorative filler.
• Editorial thumbnail generated and saved (1672×940 PNG).

Claim → Evidence Check

Official Sources

• Shopify App Store listing: https://apps.shopify.com/zipchat — checked 2026-10-07 (Zipchat AI Chatbot & Live Chat by ZIPCHAT INC; 4.9 / 204; 96/1/0/0/3%; launched 23 May 2023; Built for Shopify; Free 120 AI replies/mo ~50 conversations, 100 training pages; Starter $80 — 750 replies ~300 conversations, 1,000 pages; Pro $300 — 3,000 replies ~1,250 conversations, 75,000 pages; Scale $600 — 6,000 replies ~2,500 conversations, 300,000 pages; 7-day trials on paid; “Recurring and usage-based charges are billed every 30 days”; integrations Checkout, Shopify Admin, ChatGPT, Claude, Emails, Instagram, Messenger, Gorgias, Zendesk, Re:amaze, Klaviyo, Omnisend, Brevo, WhatsApp Business API; English)
• Shopify App Store reviews (all + 1/2/3-star filter): https://apps.shopify.com/zipchat/reviews — checked 2026-10-07
• Vendor pricing page: https://zipchat.ai/pricing — checked 2026-10-07 (Starter $64 annual / $80 monthly; Pro $240 / $300; Scale $480 / $600; Scale Plus $960 / $1,200 — 14,000 replies ~6,000 conversations, unlimited training pages, 10 agents; custom unlimited from $2,100/mo; 1/3/5/10 AI agents by tier; unlimited team members; all five channels on every tier; overage $49 per 250 replies on Starter–Scale, $49 per 500 on Scale Plus; 7-day free trial on every plan; 30-day money-back guarantee on first payment; no free tier listed)
• Vendor product documentation/marketing on data ingestion: zipchat.ai — checked 2026-10-07 (describes authenticating against the Shopify Admin API for read access to products, collections, customers, orders and policies; product catalogue synced on a schedule and order data queried live; product names, descriptions, variants, pricing and availability captured). Used as a developer claim to explain, not prove, the observed catalogue accuracy.
• Prior Blinto review used for the controlled comparison: reviews/tidio.md and research/tidio.md, published 2026-10-02 — same store, catalogue and theme.
• Category/assignment context: Customer Support & Helpdesk, Batch 3, Munna, SOP-1 42/50, P1 — src/lib/authorAssignments.ts, docs/batch-3-validation.md

Do not surface a source list inside the public article; facts are stated in-line in Blinto’s voice.

Baseline Research

• Core use case: AI sales-and-support agent for ecommerce across web chat, WhatsApp, Instagram, Messenger and email, with product recommendation, upsell, cart recovery, order tracking and returns.
• Target merchant: Shopify stores treating chat as a revenue channel, with volume to justify premium pricing.
• Current App Store rating: 4.9 (checked 2026-10-07) — highest rating in the Blinto set so far, on a comparatively small base.
• Review count: 204 (96% five / 1% four / 0% three / 0% two / 3% one). Re-check before publication if cited.
• Built for Shopify: Yes. Launch date: May 23, 2023.
• Pricing model: Reply-based tiers with block overage; premium relative to the category.
• Free plan/trial: Free tier on the listing only (120 replies/month); 7-day trial on paid plans; 30-day money-back on first payment.
• Languages: English on the listing, though multilingual answering is marketed and appears in merchant praise.

Developer Claims

Merchant Feedback Patterns

Common positives (recent reviews, Jun–Sep 2026): natural, accurate, human-like answers; strong revenue attribution, including claims of the app returning several times its cost; fast no-code setup; named onboarding and support staff praised repeatedly; reduced repetitive-question workload.

Common complaints (1–3-star sample, Feb 2024 – May 2026, checked 2026-10-07):

• Support replies exceeding a week while on a paid plan (May 2026, no reply shown).
• Support slow and ineffective; merchant unable to get past initial setup (Feb 2026, no reply shown).
• Weak performance on a niche catalogue; AI responses described as off-putting and product suggestions as random; promised custom instructions never delivered (Mar 2026, developer reply).
• No data export to CRM or phone systems; old training data persisting after retraining attempts (Nov 2025, developer reply).
• Widget still visible in the storefront footer after uninstall (Apr 2024, no reply shown).
• Unsolicited marketing email from a sales rep after the merchant switched to a competitor; the developer’s public reply defended the outreach as standard practice (Apr 2025).
• Unspecified dissatisfaction / better alternatives available (Feb 2024).

Interpretation boundary: A very small negative sample (about 3% of reviews), sourced, not reproduced. Our test did not touch support, export, retraining, uninstall or outreach. The assignment brief flagged outreach complaints as a known risk; that is confirmed as present in the public record but was not experienced first-hand.

Missing Evidence / Questions

• Which plan the test store was on, and how many of the free tier’s 120 replies the session consumed.
• Whether the Make correction control changes future answers, and what Debug exposes.
• Whether the discount behaviour actually fires, what it issues, and whether it respects Shopify discount rules.
• How the agent performs across a broader question set, especially stock, order status and returns.
• Whether answers hold on a niche catalogue, given the sourced complaint.
• Whether the storefront bar can be repositioned or restyled, and how it renders on mobile.
• Munna’s own impressions: setup difficulty, confidence in the agent, value judgement.
• ClickUp task URL.

Editorial Evaluation

Strengths

1. Correct comparative catalogue answer, verified against the storefront, on the question that defeated a rival on the same store.
2. Clarifying question on an ambiguous prompt rather than a confident guess.
3. Complete question-to-cart loop inside the chat surface at the correct price.
4. Zero-configuration catalogue learning; per-message Debug and Make correction in the merchant inbox.

Limitations

1. Premium pricing: $80 entry, $49 per 250 overage replies, free tier of 120 replies.
2. Discount-to-hesitant-buyers at up to 20% and email collection enabled by default.
3. Prominent centred storefront bar with no theme-level settings.
4. Sourced complaints: slow support on paid plans, no conversation export, stale training data, post-uninstall persistence, unsolicited outreach.

Best Fit

Shopify stores treating chat as a sales channel, with margin to absorb AI-issued discounts and volume to justify the subscription; stores with large price-varied catalogues where comparative questions are common.

Poor Fit

Budget-constrained ticket deflection; merchants needing conversation export into a CRM or helpdesk; niche catalogues where generic recommendations underperform.

Alternatives to Consider

Tidio (reviewed by Blinto on this same store); Gorgias, Re:amaze, Richpanel, DelightChat, Crisp, Chatra (Batch 3 candidates, not yet reviewed).

Claims We Must Not Make

• Do not state an accuracy rate for Zipchat or Tidio, or declare one app better; we asked a handful of questions in one session each.
• Do not assert the Admin API mechanism as the proven cause of the correct answer; it is a developer claim consistent with observation.
• Do not claim the discount behaviour works, fires correctly, or respects discount rules; only the setting was seen.
• Do not claim hand-off, take-over, correction, Debug, campaigns, reports or any non-chat channel works.
• Do not claim a checkout completed; the cart was reached, not the order.
• Do not attribute the store’s Web Vitals figures to Zipchat.
• Do not present the sourced support, export, retraining, uninstall or outreach complaints as reproduced.
• Do not repeat merchant revenue claims (“pays for itself ten times over”) as verified.
• Do not state which plan the test ran on.
• Do not describe the reviewer’s subjective experience until Munna supplies it.

Blinto Review Quality Score — Internal

Status: 90–100 — Strong, publishable. Scored on 2026-10-07; the article was published the same day with text identical to this record. At 94/100 it cleared the 81+ auto-approval threshold in AGENTS.md §10, so manual SEO Check was not required.

Editorial Verdict Lock

• Verdict locked: Yes — article published by the project owner on 2026-10-07 with text identical to this repository version.
• Merchant decision: Recommended for merchants who want chat to sell, have checked the volume maths, and review the discount defaults before going live.
• Best for: Sales-led stores with margin and volume; large price-varied catalogues.
• Not for: Budget ticket deflection; export-dependent workflows; niche catalogues.
• Verdict notes: Consistent with the supplied evidence and the eight published screenshots.

Founder Intelligence — Internal

What the app does exceptionally well

• Catalogue grounding: the agent answered a comparative pricing question correctly and produced a correct runner-up, where a competitor on the identical store and catalogue did not.
• Uncertainty handling: it asked which product the shopper meant instead of guessing.
• Commerce loop: product cards with add-to-cart put the right item in the Shopify cart without a page load.
• Zero-configuration learning — no knowledge-source field to fill in, unlike the competitor tested on this store.

Merchant friction discovered

• Three behaviours ship enabled, including a 20% discount ceiling, before the merchant has seen a conversation.
• Pricing is the steepest in the category reviewed so far, with block overage at roughly twenty cents per reply and a free tier too small to evaluate properly.
• The storefront surface is a prominent centred bar with no theme-level configuration.
• Public complaints concentrate on support latency on paid plans, absent conversation export, stale training data and post-uninstall residue — and one on sales outreach to a departed customer, which the company defended publicly.

Growth opportunities

• Ship the discount behaviour disabled, or gate it behind an explicit confirmation; it is the single setting most likely to cause regret.
• Publish the free tier on the website pricing page; it currently appears only on the Shopify listing.
• Revisit the public response to the outreach complaint — defending it in a review reply compounds the reputational cost.

Product opportunities

• Conversation export to CSV/CRM, the clearest functional gap in the negative reviews.
• Make retraining visibly supersede old data, with a timestamp the merchant can see.
• Expose what Debug shows and whether Make correction propagates, so merchants can trust the loop.

Questions for the founder

• Does the catalogue sync interval create a window where price answers can be stale?
• What exactly does Make correction change — that conversation, or future retrieval?
• How are discount offers bounded against existing Shopify discount rules and margin?

Blinto fit

Product/UX opportunity

Evidence-based reason: The core retrieval and commerce loop are the strongest we have tested in this category; the friction is in defaults that ship hot, pricing communication, and post-sale trust signals — product, packaging and positioning problems rather than engineering ones.

SEO & Content Direction

• Target merchant: Shopify merchants comparing AI chat agents on answer accuracy and sales capability, and merchants researching Zipchat pricing.
• Merchant problem: Whether an AI agent can be trusted on catalogue facts, whether it can sell, and whether the price is justified.
• Primary search intent: Commercial investigation.
• Primary keyword/topic: Zipchat review
• Secondary topics/questions: Zipchat pricing, Zipchat vs Tidio, Zipchat AI accuracy, Shopify AI chatbot that sells, Zipchat free plan.
• SERP/content gaps: The brief flagged an affiliate-dominated SERP. Competing pages restate the feature list and repeat vendor revenue claims; none runs a controlled same-store comparison or quantifies overage.
• Blinto differentiated angle: The controlled same-store, same-question comparison with the Tidio review, plus the in-chat cart loop and the discount default.
• Internal links: reviews/tidio.md (same category and same store — linked from this article; consider adding a reciprocal link from the Tidio review once a third review lands).
• CTA direction: Run the same comparative question against any agent you trial; check the reply-volume maths including overage; review the behaviour defaults before launch.
• Required disclosures: Level 1 test of web chat only; development store; same store as the Tidio review; one session; channels, campaigns, correction tooling and billing untested; pricing and review data checked 2026-10-07.

Thumbnail Concept and Generation Prompt

Concept: The deliberate counterpart to the Tidio thumbnail. A traveller at the same kind of sunlit counter; the pale attendant holds out one small object, and its reflection in the polished surface matches exactly. Reads as “asked, and answered correctly.”

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: a small lone traveller with a simple bag stands at a long polished stone counter in a vast sunlit hall with tall windows. Behind the counter, a softly glowing pale figure leans forward and holds out a single small object in an open palm, offering it directly to the traveller. The object’s reflection in the polished counter matches it exactly. The traveller is the only soft focal point and is small in the frame; the hall and counter dissolve into broad blurred masses of colour.

Light: high in the upper background, through a tall window, a single soft pastel blue ring shape glows faintly like a distant sun seen through haze — feathered edges, varied opacity, painted with the same loose brushwork, never a logo or a clean circle.

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 UI elements, no screens, no phones, no speech bubbles, 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: The ring’s tint derives from the pale ice-blue of Zipchat’s app icon (sampled from the install dialog: #BAE8FF/#C4E8FE on a #111727 navy ground), softened into a pastel against the painting’s cream light. Deliberately cooler and lighter than the Tidio thumbnail’s mid-blue ring so the two Customer Support reviews read as a pair without looking identical. Verify the exact brand hex from the live listing before generating; do not let the image tool reproduce a crisp logo.

Public Article Authenticity Gate

• What We Tested is visible in the article and accurately describes scope and limitations.
• Eight purposeful hands-on screenshots are embedded and the files exist in the repository.
• Each planned screenshot supports a merchant-relevant claim.
• Evidence is placed beside the relevant finding.
• Major first-hand findings use accurate first-hand language and map to this research record.
• Developer claims and merchant feedback are distinguished from Blinto observations.
• The accuracy result is explicitly bounded to one question in one session, twice.
• The comparison with Tidio is framed as a controlled observation, not a ranking.
• Store-level Web Vitals are excluded rather than attributed to the app.
• Selected public evidence is redacted and publication-safe.
• Requested read-time constraint met (2,677 body words; 13.4 min at 200 wpm).
• Editorial thumbnail generated and saved at the frontmatter path.
• Selected evidence renders correctly on the Review Hub and the public article.
• Project owner reviewed the rendered evidence and published the article on 2026-10-07.

Publication Gate

• Material hands-on claims are supported by supplied screenshot evidence.
• Pricing/time-sensitive facts verified 2026-10-07.
• Important contradictions are resolved or qualified (free tier on the listing but not the vendor site; mechanism offered as claim not cause; accuracy bounded).
• Unsupported claims are removed.
• Evidence is sufficient to write a useful merchant-facing verdict.
• Public Article Authenticity Gate is fully complete.
• Blinto Review Quality Score is complete and 80+ (94/100) — cleared the 81+ auto-approval threshold.
• Editorial verdict is locked.
• Founder Intelligence is recorded internally.
• Manual SEO Check not required per AGENTS.md §10. Review preview live at /reviews/zipchat/; published publicly at https://blinto.co/blog/zipchat-shopify-review/ on 2026-10-07.

The internal research preview will be /research/zipchat/. Any SEO handoff must use the rendered review URL /reviews/zipchat/, never the research URL.