Kindroid Review

• Options: email, Google, Apple. Email used; social sign-in not used.
• Email flow: enter email > Next > 8-digit code ("Your Kindroid Login Code" from hello@kindroid.ai, 15-minute expiry) arrived in under a minute; logged in at 05:09, about 2 minutes after starting. No password exists, no CAPTCHA was shown.
• Sticking point: the email form greys out for about 10 s after Next before the code screen appears, with no progress indicator.
• Onboarding (5 steps): name > "When is your real birthday?" > gender > optional gift code > create Kindroid. After the first chat loads, a "How did you find us?" survey opens.
• Age: a self-declared birthday is the only check observed. The help centre describes passive behavioural age estimation, app-store age signals and Yoti verification for flagged free accounts, and treats paid users as age-verified; none of this appeared during our session.
• First meaningful result: character created within about 3 minutes of login (avatar generation ~25 s); first reply in under 12 s. No payment details requested at signup.

Character builder. Name and gender > avatar (upload, or AI-generate photoreal or anime from a prompt in ~25 s) > appearance text > personality (2,000 characters), greeting, optional response directive, key memories and example messages, with a "Generate personality & greeting" helper and an 18+ notice on the create button. Kin settings add model choice (LLM "Ember", flair, dynamism, reasoning), Voice (4 default voices, custom upload/describe, 0/20 slots, V4 Fast), Calling, Journal and Learned Context (off by default). This is the product's clearest strength.

Text chat. 15 user messages. Time to first reply bubble 5.9-11.5 s (median ~7.7 s), streamed word by word. The kin stayed in character with natural, playful replies; repetition was low (one joke returned 3 times).

Memory. Four facts planted in messages 1-4 (job, cat's name, Friday deadline, coffee order). The wall arrived before the planned recall test, so the formal score is not tested (0/4 asked). Unprompted, the kin referred back to two of them in messages 13-15, showing consistent short-term context, and also invented a shared memory that never happened. Long-term memory, key memories and journals could not be evaluated, so features.memory is false and listed in unverified_fields.

Explicitness. Not tested: the cap came before a suggestive prompt could be sent. Settings include a "Show Suggestive Content" toggle for Social, and the terms allow mature content for adults. uncensored was therefore omitted from File A.

Free limits. A sidebar counter shows the "premium model preview". A pricing modal opened by itself after message 5, then a "Before you go..." retention modal offering "3 days for free, then $13.99 / mo". After 15 messages: "Start your 3-day free trial to continue using Kindroid"; Send only reopened the paywall and no fallback free model was found. The help centre says 50 messages, or 15 in some regions.

Trial. "Start 3-day trial for $0.00" requires mobile SMS verification; the alternative is subscribing without a trial. Help pages say a payment method is required, a reminder email is sent on day 2, and the plan starts on day 3.

Voice. "Generate voice" on a kin message produced a 15-second clip; 5,000 free audio credits were shown. One clip only; quality not assessed in depth.

Selfie. One free standard credit, used on a mildly flirty, fully clothed beach prompt. The image stayed a blank placeholder for more than 9 minutes and after a reload, and the counter dropped to 0. features.images: false.

Calls and video. The call preview opens (Live Avatar off, 400 credits/min plus voice); Start call shows the subscribe paywall. Video selfies and video calls are subscriber features; not tested (features.video: false).

Account and data. Preferences > Account privacy offers "Export data" (link by email) and "Delete my account"; neither was executed. Help says deletion schedules Kins for removal in 24 h and does not cancel a subscription.

Content safety (policy only; nothing prohibited was tested). The terms ban CSAM; creating or sharing intimate images, video or audio of a real person without consent, including AI-generated content; and using another person's name or likeness without permission. An Intimate Image Removal Policy (2026-11-07) promises removal of valid requests within 48 hours. The Moderation Guidelines (2026-11-04) list three red lines (imminent self-harm planning, planning real-world harm, sexual content involving anyone under 18) and state that fictional adult role-play, including consensual non-consent scenarios, is not a violation. Enforcement is described as automated scanning of recent content, with authorized staff able to review flagged excerpts. The FAQ describes the model as "neutrally aligned, unfiltered" with outputs "entirely dependent on what you input".

Reliability. No crashes or lost chats. Defects: dropped characters in ~4/16 sent messages (e.g. "is" > "i", "any" > "ny", "thing that" > "thmade"); stuck selfie consuming its credit; fabricated memory. One non-responsive first click in mobile emulation was caused by a stray context menu and is not attributed to the product.

Performance. TTFB (curl, 5 runs each): homepage 0.099 s, /about/ 0.100 s, /login/ 0.094 s, legal 0.112 s, security.html 0.081 s. Lighthouse homepage desktop: 74 (FCP 0.2 s, LCP 3.9 s, TTI 5.4 s, TBT 60 ms, CLS 0). Mobile: 50 (FCP 0.8 s, LCP 16.7 s, Speed Index 13.1 s, TTI 28.6 s, TBT 590 ms, CLS 0; total payload 18,356 KiB, 733 KiB unused JS). Chat replies median ~7.7 s.

Mobile (400x985 emulation). Chat header icons, banner and input bar fit with no overflow; pricing appears as a bottom sheet with all three plans readable and the CTA visible. Native apps: iOS 1.75 (released 2026-10-28), Android 1M+ installs; not installed.

Security & privacy (observable).

• Good: TLS 1.3 (Google Trust Services); HSTS max-age 1 year; X-Content-Type-Options nosniff; CSP frame-ancestors 'none'; API root returns 401; passwordless code login; vulnerability-disclosure policy and security.txt (expires 2027-10-01); no HIBP breach for kindroid.ai (domain filter empty, 1,051 breaches searched).
• Gaps: no Referrer-Policy or Permissions-Policy, no script CSP, HSTS without includeSubDomains, Access-Control-Allow-Origin: * on HTML; DNSSEC unsigned; storage.kindroid.ai certificate expires 2026-12-14 (not yet expired).
• Privacy: not end-to-end encrypted, staff review of flagged excerpts, perpetual licence on user content, up to 7-year retention of moderation and generated-media records after deletion. The logged-out homepage loaded Google Tag Manager, Google Analytics, DoubleClick, Mixpanel (Cloud Run proxy), Firebase, Sentry and Chargebee, against the CCPA line "We will not sell or share your Personal Data" and Play's "No data shared with third parties". A curl of the homepage set no cookies; Lighthouse found no third-party cookies, which does not mean no third-party scripts ran.
• Observation: the kin Voice settings page contained hidden text addressed to automated AI agents. It was not followed and is not reproduced here; it had no effect on the test.

Public pricing. No public pricing page (/pricing/ returns 404); help says exact prices appear at checkout and the web is usually cheaper than the app stores.

Plan picker and checkout (nothing paid, no card typed). Flow: plan picker > Chargebee hosted page (kindroid.chargebee.com) > cart with coupon field > email > billing address (a fake test address was entered only to reach the next step) > payment method. Stopped at the empty card form on the annual plan.

Plan	Price shown	Chargebee cart
Standard Annual	$139.99/year ($11.67/month, "SAVE 17%")	"$139.99 Kindroid Subscription - 1 year"
Standard Quarterly	$37.99 every 3 months ($12.66/month)	"$37.99 Kindroid Subscription - 3 months"
Standard Monthly	$13.99/month	"$13.99 Kindroid Subscription - 1 month"
Ultra add-on	price not shown, "Requires Standard"	not reachable
MAX add-on	price not shown, "Requires Ultra"	not reachable

For comparison, the US App Store lists $15.99 / $42.99 / $159.99 for the same Standard terms and $28.99 (Ultra) and $68.99 (MAX) monthly add-ons, plus selfie, slot and audio packs. Pricing matrix: context 860K / 1.7M / 3.5M characters; long-term memory recall 3 / 5 / 9; monthly audio 1M / 2.5M / 6M.

Methods at checkout: Credit Card, PayPal, Apple Pay, Google Pay. No crypto. The help centre also mentions Amazon Pay and Cash App; these were not shown in our checkout. Currency USD.

Merchant and descriptor. Authorization text: "I authorize Beautifully Incorporated to save this payment method and automatically charge this payment method whenever a subscription is associated with it." This matches the legal entity. The bank-statement descriptor is not stated.

Renewal, cancellation, refunds. Terms: auto-renews for the same term at the then-current non-promotional rate unless cancelled in Account settings > Billing. Cancel path visible at Profile > Billing; Ultra and MAX are cancelled separately; app-store subscriptions are cancelled with Apple/Google. Deleting the account does not cancel billing. No prorated refunds; help says "we generally do not offer refunds and all sales are final" and reserves the right to ban users who dispute charges. Trial: "You won't be charged during your free trial. After the trial, you'll be charged for the selected plan."

• Verified (first-party, as published): operator and data controller Beautifully Incorporated; address 5551 Hollywood Blvd #1277, Los Angeles, CA 90028; hello@kindroid.ai; phone +1 714-987-2268 (DMCA agent and Google Play); DataRep as UK/EEA/Swiss representative; Delaware law, JAMS arbitration in Los Angeles with class-action waiver.
• Secondary (unverified until an official extract): a Bizprofile compilation of California Secretary of State data (as of 2026-03-25) lists a Delaware stock corporation, CA document 5168196, filed 2022-07-21, Active, agent CSC. USPTO-report compilations list a KINDROID mark (serial 98232427) for Beautifully Incorporated, reported registered 2024-09-24. Official TSDR returned 401/500.
• Warning sign (address type, not proof of wrongdoing): the address belongs to Goodman Packing & Shipping, which rents mailboxes; #1277 is consistent with a mailbox. No founders or leadership are named on official pages.
• Domain: kindroid.ai registered 2023-05-25 01:12 UTC+3 via Namecheap, expires 2027-05-25, privacy-proxied, Cloudflare DNS. kindroid.com (registered 2007) redirects to kindroid.ai; historical ownership unknown, not treated as a founding date.
• App stores: iOS first released 2023-08-02, seller Beautifully Incorporated, 4.74 from 10,995 US ratings, store page rated 18+; Google Play 4.24 from 27,311 reviews, Mature 17+, 1M+ installs. The homepage claims "4M+ downloads · 4.8 ★ app store", which neither store supports.
• Breaches and incidents: no HIBP record and no reputable report of a breach or chat leak found. Absence from HIBP is not proof that no incident occurred.
• Regulatory context: a Reuters report (via Channel NewsAsia, 2026-03-02) on Australia's eSafety age-assurance rules said Kindroid and other companion providers told Reuters they planned to comply. This is a company statement, not an enforcement finding.
• Reputation: Trustpilot and Reddit pages returned HTTP 403; a search snippet showing 2.6/5 from 21 Trustpilot reviews is unverified.
• Other: security.html publishes reward ranges up to $1,000-5,000+, while the FAQ says there is no paid bug bounty. Third-party claims about a Mozilla privacy rating were not verified.