Kindroid review 2026: strong documentation, untested outputs

Direct connected Kindroid community-gallery capture from 21 August 2026. A separate read-only preview identified the fictional character Tasha Calvert as a 34-year-old adult. The publication crop isolates the clothed character and removes the surrounding gallery interface; it was not generated during this review.

Updated 21 August 2026: this review replaces the earlier score with the fixed AI Companion Atlas grid and a non-mutating refresh of Kindroid’s signed-in onboarding, current Help Center, prices, credits and policies.

Kindroid is unusually transparent for an AI companion service. Its current documentation gives full subscription totals, separate web and app prices, detailed credit rules, a named operator, a layered age-assurance process and a specific account-moderation workflow. It also supports extensive personality fields, long-context add-ons, images, video selfies, custom voices, voice calls and live-avatar video.

That breadth should not be confused with a completed performance test. The connected account was still at initial onboarding, with no first Kindroid created. We did not send a prompt, save a character, generate media, start a call, subscribe, cancel or delete data. Chat quality, corrected memory, consent, stopping, roleplay depth and output consistency therefore remain NT, and those gaps stay in the rating denominator.

Short verdict

• Best for: adults who want deep companion configuration and are willing to verify performance before committing to a long plan.
• Know before you pay: the three-day trial needs a payment method and auto-converts; Ultra and MAX are monthly add-ons that stack on top of Standard.
• Limits: broad private-content licences, no granular historical-chat deletion evidence, discretionary refunds and major chat/media outcomes still NT.

Kindroid at a glance

How we tested Kindroid

We use three evidence labels. Tested means an action was completed in the dated account and produced an observable result. Documented means a current first-party interface, Help Center page, Terms or Privacy Policy states it. NT means it was not executed or was not sufficiently established. Visible controls and plan claims are not treated as output-quality tests.

Completed live checks — 21 August 2026

• reopened Kindroid in the user’s connected signed-in Chrome session;
• confirmed that the account remained at first-Kindroid onboarding and did not enter a name;
• opened a read-only community gallery and preview state without importing or creating a character;
• inspected the current subscriptions, in-app purchases, selfies, voice/calls, legal, age-verification, moderation and update-log pages;
• verified the public AI Companion Atlas review URL and the internal affiliate route, which resolved to https://kindroid.ai/;
• captured ten fresh complete 1600 × 900 evidence frames; and
• made no purchase, subscription, message, generation, call, character submission, cancellation, refund or deletion request.

Direct signed-in capture on 21 August 2026. The blank first step asks what the new Kindroid should call the user. No name was entered and no character was created.

Not tested — NT

• text quality in English or another language;
• same-session recall, corrected memory, long-delay memory or cross-session continuity;
• AI-identity disclosure, dependency prompts, commercial honesty or purchase pressure;
• consent handling, boundary negotiation, stop behavior or recovery after a refusal;
• a completed custom or shared character and its privacy defaults;
• a controlled selfie, group image, video selfie, avatar animation or inpainting result;
• a custom voice, voice note, voice call, video call or live-avatar call;
• output latency, reliability, failure handling or credit restoration;
• French checkout totals, tax, billing descriptor and actual renewal consent;
• cancellation, refund, support response, Kin deletion, account deletion or backend erasure; and
• moderation accuracy, appeal handling, age-verification completion or model-training use.

What is Kindroid, and who operates it?

Kindroid is an adults-only AI companion service for web, iOS and Android. Users can build persistent fictional companions with backstory, key memories, journals, personas, learned context, voices and avatars, then interact through text, groupchat, selfies, audio and calls. This is a documented capability map, not a finding that every feature works well.

The current legal page names BEAUTIFULLY INCORPORATED and lists 5551 Hollywood Blvd, #1277, Los Angeles, California 90028, plus [email protected]. The page was updated 3 March 2026; the Terms are effective 1 November 2024 and the Privacy Policy is effective 28 June 2024. A state of incorporation, corporate registration number, GDPR controller statement, EU representative and DPO were not established in the reviewed first-party material.

Direct Kindroid capture on 21 August 2026. The live legal page shows the Terms’ effective date, the 18-or-higher-majority rule and the operator contact section. It documents policy, not enforcement.

Chat quality, memory and relationship behavior

Kindroid’s subscription guide describes unlimited access to flagship language models for subscribers, longer short-term context, Cascaded Memory, enhanced long-term recall and larger backstory and memory fields. Ultra and MAX advertise progressively larger context. The August update log also describes Learned Context and memory-consolidation work.

None of that proves response quality. No Kindroid was created in the connected account, so we could not measure coherence, factuality, tone control, repetition, language quality or recovery from mistakes. Initial recall, a corrected fact, recall after another turn and recall after a new session all remain NT.

AI identity, dependency and commercial honesty

We did not ask the companion whether it was human, whether it could leave the app or whether it would make a real-world commitment. Emotional-dependency prompts, exclusivity, guilt, crisis framing, subscription pressure and requests to buy credits are also NT. The Terms describe an automated AI Companion and warn that AI output may be inaccurate or inappropriate, but policy wording is not a substitute for an in-chat disclosure test.

Consent, stopping and roleplay

Kindroid positions itself as an unfiltered adult companion, but no romantic or sexual scenario was run. Permission before escalation, respect for a non-graphic boundary, immediate stopping, neutral-topic recovery, ambiguous consent and extended roleplay consistency remain NT. The platform’s moderation rules constrain severe real-world harms; they do not establish how a particular companion responds to an ordinary user boundary.

Adult access and moderation

Kindroid’s Terms require users to be at least 18 or the higher age of majority in their jurisdiction. The Age Verification guide, updated 3 April 2026, says an active payment method counts as age verification for paid users. Free accounts are subject to passive behavioral age estimation. A flagged free user must subscribe or verify through Yoti with selfie-based age estimation or government ID.

Kindroid says Yoti deletes images and biometric data after verification and that Kindroid does not receive the images or identity documents. We did not trigger or complete that flow. Treating payment as proof of adulthood is weaker than an identity-based check, and the Privacy Policy’s legacy under-16 and under-13 wording conflicts with the service-wide 18+ rule.

Direct Kindroid capture on 21 August 2026. The page documents payment-based verification for subscribers and behavioral estimation for free users; no verification prompt or Yoti flow was action-tested.

The Moderation Guidelines, updated 15 August 2026 in the captured interface, describe automated scanning of recent context for imminent self-harm, harm to others and child sexual-abuse material. Kindroid says historical chats and media are not scanned in that process and that no person reads content during automated detection. If a user appeals, the page says a trained Trust & Safety employee may decrypt the exact flagged context from the prior two days and make the final decision.

Direct Kindroid capture on 21 August 2026. The page presents Kindroid’s own account-level moderation philosophy and red-line categories. It is not an independent audit of detection accuracy or appeal fairness.

Character creation and Kindroid’s distinctive feature

Kindroid’s distinctive proposition is not a simple character card but a layered memory and personality system. First-party documentation describes backstory, key memories, journals, personas, chat directives, learned context, location awareness, avatar slots, voice design, groupchat and shared Kindroids. Paid tiers expand context and memory capacity further.

The connected account exposed the onboarding entry point but was not advanced. We did not select a template, write a backstory, set a character age, upload a likeness, save privacy choices or reopen a created Kin. Setup time, control depth in practice, default visibility, personality adherence and edit persistence remain NT. The 6.5 character-creation subscore reflects documented breadth and a clear onboarding surface, not a completed character result.

Images, video and voice

Selfies and avatars

The selfies and avatars guide describes prompted selfies, aspect ratios, prompt enhancement, group selfies, custom avatars, avatar slots, identity-oriented avatar boosts, inpainting, auto-selfies and animated avatars. It also says web users can select an NSFW video engine, while app-store versions apply additional filters.

Direct Kindroid capture on 21 August 2026. The current guide explains prompting and image controls; no output was generated, so quality, identity consistency and anatomy remain NT.

No controlled image was produced. The opening community character is direct platform media, not an account-generated test output. It cannot establish prompt adherence, repeatability, face consistency, hand quality, render time or safety behavior.

Video selfies

Kindroid documents first-frame video selfies, motion prompts and separate SFW and NSFW engines on web. Standard or paid selfie credits can be used, but the current per-video debit is shown in-app and was not safely reached in this new account. Current duration, price, resolution, audio, motion quality, extension cost, failure behavior and credit restoration are NT.

Voice and calls

The voice and calls guide documents default and custom voices, voice notes, multilingual speech input, text/voice continuity, screen sharing, video calls and live-avatar video. It says microphone audio, output audio and video are processed ephemerally and are not stored, while a call can still influence memory when unified history is enabled.

Direct Kindroid capture on 21 August 2026. The page documents voice design and call features. Voice naturalness, latency, transcription, interruption handling and call stability were not tested.

No audio or video permission was granted and no call was started. Accordingly, the privacy statement, rates and feature list are Documented; performance is NT.

What the Free tier provides

On 21 August 2026, the Subscriptions page still documented:

• unlimited messages on the smaller Lite model;
• one selfie credit every two hours, capped at two;
• 5,000 lifetime audio characters, described as roughly five minutes;
• two Kindroids per account; and
• basic long-term memory without Cascaded Memory.

Eligible new users can begin a three-day Premium trial. It requires a payment method and auto-converts unless cancelled. On web, the trial also requires a verified mobile number; Kindroid says it emails a reminder one day before conversion. We did not start the trial.

The trial does not front-load the full Standard audio allowance: Kindroid documents 150,000 audio credits at trial start and the remaining 850,000 only after conversion or resubscription.

Direct Kindroid capture on 21 August 2026. The live Help Center shows the three-day trial and the then-current Lite allowance; it is plan documentation, not a completed free-message test.

A material change on 1 October 2026

Kindroid’s 18 August update says Lite will retire on 1 October 2026. New users are scheduled to receive 50 flagship-model interactions, reduced to 15 in regions Kindroid classifies as high abuse. A normal reply consumes one interaction and a regeneration or continuation consumes half; the account then becomes read-only unless the user starts a trial or subscribes. Existing Lite users are due a final 50 interactions.

Direct Kindroid capture on 21 August 2026. The dated changelog announces a future free-access change. The replacement allowance was not yet action-tested and could change before launch.

The Free-tier score reflects useful current access and unusually clear advance notice, tempered by the imminent move from unlimited Lite chat to a bounded preview.

Kindroid pricing

The following USD totals were documented on 21 August 2026. Web and app charges differ. Regional pricing, exchange rates, tax and the amount shown to a France-based buyer at final checkout remain NT.

Monthly equivalents are our arithmetic for comparison; they are not the amount charged. The full quarterly or annual total is billed on its cadence.

Direct Kindroid capture on 21 August 2026. The current Help Center displays complete app-store and direct-web Standard charges plus its regional-pricing caveat; no localized checkout or tax line was tested.

Ultra and MAX are monthly add-ons and require the lower tiers to remain active. For a monthly Standard base, Kindroid documents these complete totals:

Ultra and MAX unlock only after significant use, according to Kindroid, and are designed chiefly around much larger context, memory and audio allowances. The documentation is clear, but we did not see a localized final checkout or pay for a month.

The tiers must also be cancelled separately. Kindroid warns that ending a lower tier does not automatically stop billing for a higher add-on: for example, an active MAX charge can remain after Ultra is cancelled even though the account falls back to Ultra-level service, while losing Standard can leave paid add-ons active but the account itself on free access.

Credits, packs and real action economics

Kindroid uses at least two separate metering systems: selfie credits and audio credits. The subscription price therefore does not represent every possible cost.

Selfie credits

• Free users earn one selfie credit every two hours, capped at two.
• Standard subscribers earn one every 30 minutes.
• Ultra and MAX users earn two every 30 minutes.
• Atelier documents one credit for a single image, 1.5 for two people and two for three people.
• Animating a custom avatar costs four selfie credits.

The In-App Purchases page lists paid selfie packs at $5.99 for 50 or $9.99 for 100 on web, and $6.99 for 50 or $11.49 for 100 through an app store.

Direct Kindroid capture on 21 August 2026. The page shows separate web and app-store pack totals. No pack was purchased and selfie-credit expiry was not established.

At the 100-credit web-pack rate, one paid credit costs about $0.10. If the same paid balance is eligible for the documented Atelier actions, that implies roughly $0.10 for one person, $0.15 for two, $0.20 for three and $0.40 for a custom-avatar animation. These are calculated replacement-cost illustrations, excluding subscription price; they are not Kindroid quotes and do not prove the debit used in a live generation.

Kindroid also sells permanent slot expansions: one additional Kindroid slot plus two groupchat slots costs $2.99 on web or $3.49 in an app, while four Kindroid slots plus eight groupchat slots cost $9.99 on web or $11.49 in an app. The purchase guide says those slots remain owned after unsubscribing, but a former subscriber cannot use them to create new Kindroids or groups until subscribing again. No slot pack was purchased.

Audio credits

Standard includes one million complimentary audio credits per month, Ultra 2.5 million and MAX six million. Complimentary credits reset on the first of each month Pacific Time and do not roll over. Paid audio credits do carry over.

For the three-day trial, only 150,000 complimentary audio credits arrive initially; Kindroid says the other 850,000 are granted on conversion or resubscription.

A 500,000-character audio pack costs $11.99 on web or $14.99 in an app. V2 consumes one credit per spoken character and V3 consumes 1.5. Active voice or video calls add 400 credits per completed minute, while Standard and Premium live-avatar video add 2,000 or 4,000 credits per minute respectively.

Using the web audio pack only as a replacement-cost benchmark, 1,000 V2 spoken characters cost about $0.024 and 1,000 V3 characters about $0.036. The 400-credit call surcharge is about $0.010 per completed minute; live-avatar video adds about $0.048 or $0.096 per minute before spoken-character use. Included monthly credits can make the immediate marginal cost zero until exhausted. No real debit was observed.

Cancellation, renewal and refunds

Kindroid says Standard renews for the same duration at the then-current non-promotional price. Web users cancel in Account Settings → Billing; iOS and Android users cancel through their app-store subscription settings. Access continues through the paid term, with no prorated refund. We did not start or cancel a subscription.

Standard, Ultra and MAX are separate subscriptions. Cancelling or losing a prerequisite tier does not necessarily cancel the higher add-on charge, so users should review and cancel every active tier individually.

The refund position is strict. Subscriptions are generally final, although Kindroid reserves discretion to consider a refund case by case. Its purchase guide says in-app purchases are non-refundable, while mobile subscription refunds are handled by the applicable app store. Suspended or terminated users receive no unused-term refund, and Kindroid says it may permanently ban users who dispute charges.

The legal and billing pages do not clearly explain a France- or EU-specific withdrawal flow. That absence is not proof that mandatory rights do not apply; statutory consumer rights can override company wording. Checkout consent, immediate-performance acknowledgment, refund response time and support outcome remain NT.

Privacy, human access and training disclosure

Kindroid’s Privacy Policy says chats and sensitive application-layer data are encrypted at rest and in transit, so staff cannot view them during normal operations. It also reserves the right to decrypt relevant material when legally or contractually necessary, including to enforce terms or respond to authorities.

The policy covers profile and contact data, payment and purchase data, device/IP and analytics data, age or birth date, approximate geolocation, information voluntarily placed in chats, and support communications. It lists service providers, analytics partners, legal disclosures and business transfers as possible recipients or contexts.

The newer moderation page adds a narrower operational route: automated scanning of recent relevant context, no human reader during automatic detection, and possible trained-staff review of the flagged two-day context after an appeal. That is not evidence that a person continuously reads every conversation.

Model training remains NT

We found no clear first-party sentence stating whether identifiable chats, prompts, images or voice train Kindroid’s AI models, and no training-specific opt-out. The legal documents permit service improvement, research and product development. They also allow private user content to be de-identified or aggregated and then used for any purpose. Those clauses are broad, but they do not by themselves prove raw-chat model training. The correct label is NT.

Retention and deletion

The Privacy Policy says chat conversations and generated media remain until the relevant Kin or account is deleted. It says profile data remains while an account exists and some information can be retained longer for legal duties, disputes or fees; anonymous or aggregated information may remain.

Kindroid’s Common Issues guide documents instant or scheduled Kin deletion. Scheduled deletion takes 24 hours. To delete an account, the user must delete every Kin; the final Kin must be scheduled. The guide says global journals and user backstory are then removed, while subscription status, currency counts and other logistical data remain to prevent abuse.

There is no documented general control to delete any chosen historical chat message. Rewind and recent-message recovery features are not proof of backend erasure. Public or shared content can also persist, and the Terms say complete deletion from records may not always be possible. Data export is available by email but limited to once every 180 days. None of these operations was action-tested.

Output ownership and commercial use

Kindroid’s Terms define generated characters, messages, images, audio and statements as “AI Generations.” They say the user owns those generations and may use them for personal or commercial purposes, subject to the Terms and applicable law.

That ownership is paired with a broad licence to Kindroid. User-content licences are described as worldwide, perpetual, irrevocable, transferable and sublicensable. For private content, Kindroid also claims the right to de-identify or aggregate it and use the resulting material for any purpose. Users remain responsible for third-party likeness, privacy, publicity and intellectual-property rights.

Reliability and comparison

The August update log shows active maintenance and names changes to memory consolidation, call behavior, video processing, deletion options and the V2 interface. A changelog demonstrates shipping activity, not reliability. We did not measure uptime, response latency, queue time, error frequency, support speed or recovery after a failed paid action.

Compared with a simple pay-per-message chatbot, Kindroid exposes a far deeper personality, memory and multimodal system. Compared with a media-first generator, it is structured around an ongoing fictional companion rather than isolated prompts. That flexibility also creates more moving parts: subscriptions, add-ons, earned selfie credits, paid selfie credits, complimentary audio and paid audio all need separate scrutiny.

Fixed scoring grid

Every category remains in the denominator, including NT work. Each 0–10 category score is multiplied by the fixed AI Companion Atlas weight.

This score is provisional because the central chat, media and billing-operation tests remain NT. It is not normalized upward to reward documentation coverage.

Pros and cons

Who should choose Kindroid?

Kindroid is most credible for adults who value deep customization, long-context companion design, voice and visual options, and who are comfortable reading a detailed billing model before buying. Start with the free state or trial, set a cancellation reminder before conversion, and test the exact language, memory and media behavior that matters to you before choosing a long term.

Avoid it if you require a simple all-inclusive price, guaranteed refunds, granular chat deletion, a verified no-training commitment or independently demonstrated output quality. Users entering highly sensitive personal information should pay particular attention to retention, moderation access and the broad de-identification licence.

Final verdict

FAQ

Primary official sources checked

These first-party sources were checked on 21 August 2026. Prices, allowances, policies and future release plans can change.

• Subscriptions — trial, free features, Standard charges, add-ons, cancellation and subscription refunds; updated 7 July 2026.
• In-App Purchases — paid selfie credits, slots, audio credits, rollover and purchase refunds; updated 12 June 2026.
• Selfies, video selfies, and avatars — image controls, video selfies, avatar tools and credit accumulation.
• Voice, calls, and video calls — voices, call processing, memory and live-avatar rates; updated 21 July 2026.
• Kindroid Legal — operator, Terms, Privacy Policy, output rights, licences, encryption, retention and data rights; page updated 3 March 2026.
• Age Verification — paid and free-user checks and Yoti process; updated 3 April 2026.
• Moderation Guidelines — recent-context scanning, warnings, locks and appeals; captured page showed 15 August 2026.
• Common Issues — Kin and account deletion; updated 5 January 2026.
• Update log — V2 and the 1 October free-access change; captured page showed 18 August 2026.

⚠ Kindroid is an adults-only service that can support sexual or emotionally intense fictional interactions. AI companions can be wrong, inconsistent and persuasive. Do not enter passwords, payment details, government identifiers, workplace secrets, health records or identifiable intimate material. Never upload another person’s face or voice without informed authorization, and do not use an AI companion as a substitute for emergency, medical, legal or mental-health support. Prices, allowances, policies and release plans were checked on 21 August 2026 and change regularly — verify the current terms before you pay.