Questions are part of understanding.

CatMind.ai is an experimental platform for analysing observable cat behaviour in short videos. It combines video frames, audio events, movement, posture, visible anatomical cues, scene context and information from your cat's profile to produce a structured, evidence-aware interpretation.

No. CatMind does not convert a meow into a literal sentence and does not claim to read a cat's mind. It is designed as a serious analysis tool that separates measured observations from possible explanations.

The creators have worked to make CatMind as understandable as possible, but the platform performs a complex analysis of serious data. Terms such as confidence, pose landmarks, tracking, calibration, hypothesis and evidence gate are retained when they make the result more transparent. Plain-language summaries appear first, with technical detail available for users who want to inspect it.

CatMind is currently designed and tuned for domestic cats. Results for dogs, wildlife or other animals should not be considered supported.

No. CatMind provides experimental observations and possible interpretations of audio-visual behaviour. It does not diagnose illness, pain or injury and must not replace a veterinarian or qualified animal behaviour professional. If you are concerned about your cat's health or safety, contact a veterinarian.

The CatMind Knowledge Hub contains concise, source-checked guides about everyday cat behaviour, communication, health and life stages, recording useful videos, reading confidence and understanding the limits of a single observation. It is a good first stop when your question is broader than one CatMind analysis.

No. Knowledge Hub articles are public and free to read. They explain common questions and point to relevant sources, but they do not diagnose an individual cat or replace veterinary care.

To analyse your own cat's video, yes. An account keeps each pet profile, analysis and confirmation associated with the correct user. New users can create one cat profile and try three complete analyses within 14 days without entering a payment card. The public sample below is available without an account.

Yes. Explore a real CatMind analysis of Simba, including the original video, observed evidence, interpretation, confidence and limitations. The public sample is available without registration.

You can add the cat's name, breed, sex, age, weight, height, length, neuter status, profile photo and relevant notes. Accurate profile information gives CatMind useful personal context, but a profile photo is optional.

1. Create or open your cat's profile in the Dashboard.
2. Select Upload video for analysis.
3. Choose a supported video from your device.
4. Keep the page open while the upload completes.
5. CatMind will show queued and processing progress, then open the result when it is ready.

Yes, depending on your plan. Basic supports one cat, Advanced supports two and Premier supports five. The free trial starts with one profile. See the Pricing page for the current structure.

Return to the Login page and sign in again with the email and password connected to your account. On a shared device, always log out when you finish.

Supported formats are MP4, MOV, M4V, AVI, MKV and WebM. Standard phone videos usually work best. ProRes, RAW, 8K and slow-motion recordings are not recommended.

A video can be up to 500 MB and 60 seconds. For the clearest and fastest result, CatMind recommends a natural 10–30 second clip recorded at 1080p. A 4K video may take longer to upload and process; 8K is not supported.

• Record one main natural behaviour in a focused 10–30 second clip.
• Keep the cat clearly visible, including the body, paws, tail and head whenever possible.
• Start 1–2 seconds before the behaviour and finish just after it ends.
• Hold the phone steadily and follow slowly only when the cat moves out of view.
• Use normal 1080p video, normal speed and ordinary, even lighting.
• Keep useful natural sound when possible, without television, music or conversation over it.
• Avoid digital zoom, filters, slow motion, abrupt camera movement, heavy blur and obstruction.
• Never provoke fear, distress, aggression, pain or vocalisation for a recording.
• If the cat avoids contact or tries to leave, record from farther away or stop the attempt.

See the complete Recording Guide with good, moderate and poor video examples.

The result may be weaker when the cat is partly outside the frame, only the face is visible, the camera moves abruptly, or the scene is dark, blurred or obstructed. Digital zoom and slow motion can also alter useful visual evidence. CatMind may still analyse the available signals, but missing body posture, paws, tail or natural movement can reduce confidence or lead to an insufficient-data result.

Compare the three ten-second recording examples to see how framing, stability, focus and light affect the available evidence.

Prefer one main behaviour per video. A short, complete moment is easier to interpret than a long compilation of unrelated events. If you want to compare different behaviours, record separate natural moments rather than creating several “different” examples by repeatedly cropping the same source clip.

No. CatMind can still examine visible movement, posture and scene evidence when a video has no audio stream. However, a clear natural sound track may add useful vocal and environmental evidence when sound is relevant to the moment. Keep natural cat vocalisations when possible, but avoid television, music and conversations that can hide or imitate relevant audio events.

Try to avoid recording other people, faces, private conversations, screens, documents, addresses or unrelated household details. If a person’s interaction is important to the behaviour, frame only the relevant hand when possible. If another person appears or can be heard, you are responsible for having the permission or other lawful basis required to record and upload the material.

The file may exceed 500 MB or 60 seconds, use an unsupported container, exceed the supported resolution, contain an unreadable video stream, or be damaged. Trim the clip, export a standard MP4 at 1080p and try again. Videos above 200 MB may be slow on a mobile connection, so Wi-Fi is recommended.

CatMind extracts audio and multiple groups of frames, then runs several analysis stages. Processing time varies with video length, resolution, server demand and the evidence available. Keep the page open while uploading; once the upload is complete, the job can continue through the queue.

Only one active analysis can run for the same cat at a time. Wait until the current job completes or fails before starting another upload for that profile.

Depending on the recording, CatMind can examine visual quality, cat detection and tracking, movement over time, posture and pose landmarks, visible head, ear, mouth and tail proxies, nearby scene objects, audio events and cat-like vocalisation evidence. These signals are combined with profile context rather than treated as isolated answers.

No single accuracy percentage applies to every clip or every part of the pipeline. Performance changes with lighting, angle, distance, occlusion, camera movement, background noise and how clearly the behaviour is captured. CatMind is intentionally cautious: it can lower confidence or return insufficient data instead of forcing a specific interpretation.

Confidence describes how strongly the available visible or audible evidence supports an observation or candidate within that part of the system. It is not certainty that CatMind knows what your cat intends, feels, wants or needs. Some scores are heuristic evidence measures rather than statistically calibrated probabilities; the result labels this distinction where relevant.

Read the complete guide to understanding confidence.

An observation describes something measured in the recording, such as movement level, a tracked posture or an audio event. A hypothesis is a possible explanation that may fit several observations. CatMind presents alternatives and limitations because different causes can produce similar visible behaviour.

This section goes beyond describing the visible behaviour and offers ranked, everyday reasons that could explain it. The first reason is the best fit among the possibilities supported by the available video, audio and context. It is a careful hypothesis, not proof that CatMind knows exactly what your cat is thinking, feeling or intending.

The same behaviour can have several ordinary explanations. Grooming, approaching, moving away or vocalising may mean different things in different situations. CatMind ranks the possibilities that fit the evidence and keeps alternatives visible when the recording cannot safely distinguish between them.

Use them as practical ideas to compare with what you already know about the moment and your cat's normal routine. They are not diagnoses or instructions to provoke the behaviour again. A safety note explains when repeated, unusual or concerning behaviour may be worth monitoring or discussing with a veterinarian or qualified behaviour professional.

This happens when the cat is not visible for long enough, the evidence is weak or conflicting, audio cannot be connected safely to the visible cat, or quality checks are not passed. An abstention is a safety feature, not an error. Try a shorter, steadier and better-lit recording with the relevant behaviour clearly in frame.

Each result is based on the evidence in that specific clip. Camera position, lighting, sound, objects, movement, duration and context can all change. A later model update may also change thresholds or interpretation. CatMind should be used to notice patterns across moments, not to treat one clip as a permanent label.

Learn how baseline, sequence and repeated natural episodes add context.

They are scientific visualisations of selected processing stages. Detection boxes show located objects, tracking follows the primary cat across frames, and pose or face layers show estimated anatomical points or regions. They help explain the evidence but are not perfect measurements of anatomy, emotion or health.

A Shareable Report is a focused version of one completed analysis that you can give to a behaviourist, veterinarian or trusted caregiver. It turns the result into a clear record for discussion without giving anyone access to your CatMind account or original video.

Open the latest completed analysis on the Pet page and choose Share with a behaviourist. You can create a private link that remains active for 7 days, or download the selected report as a PDF. The PDF can be saved, emailed through a service you choose or printed like any other document. A recipient opening an active private link can also download its report as a PDF.

You control whether the report includes your cat’s name, one compressed key frame, CatMind’s interpretation, scientific evidence and your personal note. Personal notes are off by default. If you omit the pet name, the report identifies the subject as “Anonymous cat”.

The original video, audio file, original filename, your name, your email, other pets and other account information are never included. A report contains only the fields you select. Check personal notes before sharing because they may contain information you entered yourself.

Yes. The Pet page shows whether a link is active, expired or revoked, its view count, when it was last opened and recent access times. Links expire automatically after 7 days, and you can revoke an active link at any time. Revocation immediately prevents further access, so create a new link if you later want to share the report again.

Personal History organises the episodes you chose to upload for one cat. It can show saved analyses, confirmed behaviours and a behaviour-over-time summary, and lets you compare two recorded episodes side by side. It is a record of selected moments, not continuous monitoring and not proof of a medical trend or cause.

Your cat’s profile and repeated usable confirmations can provide personal context for later analyses. Current video and audio evidence still comes first: past history never overrides weak, missing or conflicting evidence in a new recording, and CatMind can still return an insufficient-data result.

It lets you confirm details that CatMind may not be able to know from the recording alone: the visible behaviour, the situation, what happened just before, a reason you know or suspect, and what happened next. Usable confirmations build a private behavioural memory for that cat and can help CatMind recognise similar patterns over time.

Not automatically, and accuracy cannot be guaranteed. Personalisation begins only after enough usable confirmations are available, currently at least three. A past episode can contribute only when a new recording contains sufficiently similar, supporting evidence; unrelated or conflicting history is not applied. Clear, honest confirmations are more useful than simply submitting more answers.

Choose Not sure instead of guessing. Confirm only what you actually observed or know from the situation. A reliable behaviour, context or outcome is useful even when the underlying reason remains unknown.

No. Your confirmation records what actually happened; it does not rewrite the completed analysis. Repeated confirmed behaviour and context can support later analyses of similar episodes. Confirmed reasons and outcomes are saved for future causal personalisation, but they do not currently retrain the general CatMind model or automatically rewrite the reasons shown in Why might this be happening?

Yes. On the Pet page you can open saved analysis details, filter the list, select two episodes for comparison and delete entries. Older items may retain only a compact scientific summary rather than the original media. Deleting a confirmed episode also removes it from that cat’s personalised learning history, and deleted history cannot be restored.

The Community Showcase is a public gallery of selected CatMind stories contributed by their owners. Each approved story presents a cat's basic profile, CatMind answer, one layered evidence frame and the contextual Why might this be happening? analysis. Cards are reviewed before publication and shown in a fresh random order rather than ranked by pet name or publication date.

Open a cat's Pet page after a completed analysis and find the Community Showcase panel. Select Submit & Earn, inspect the complete public preview, confirm the separate Showcase consent and submit it for review. Nothing is submitted merely because you opened the preview.

No. A submission first receives pending status and remains private while a CatMind administrator reviews it. Only an approved contribution appears in the public gallery. A rejected contribution remains private, shows the rejection reason on your Pet page and earns no credit.

The public card uses a strict allowlist: your cat's name, breed, gender, age and neuter status when those profile fields are available; the displayed CatMind answer and confidence label; one compressed key frame with available detection, pose and face-metric overlays; and the displayed contextual hypotheses, clarifying question, suggested next step and relevant red flags. The preview on the Pet page shows what will be submitted before you consent.

Showcase never stores or publishes the original video or audio and does not extend the original media retention period. Your name, email, account ID, location, pet notes, original filename, other pets and other account information are not public. Only the approved Showcase snapshot is available to gallery visitors.

An approved contribution earns one promotional analysis credit. The credit is awarded only after administrator approval, never merely for submitting a card, and it is recorded once for that contribution. Repeated approval events or resubmitting the same analysis cannot create duplicate rewards.

Yes. Open the Showcase panel for that analysis on the Pet page and choose Withdraw submission. An approved card stops appearing publicly and its stored Showcase evidence frame is removed. A credit already awarded for the approval remains recorded, but withdrawing and resubmitting the same analysis does not earn another credit.

No. These are separate choices. A Shareable Report is a private, time-limited link or PDF for people you choose. Dataset contribution is an optional pseudonymous export used to improve CatMind. Showcase consent specifically permits publication of the previewed pet profile, evidence frame and analysis after moderation; choosing one feature does not opt you into either of the others.

The Showcase panel is offered for a completed analysis displayed on the Pet page. Submission does not guarantee publication: the preview must be available, explicit consent is required and the contribution must pass administrator review. A card may be rejected, for example, when its evidence frame is not suitable for a clear public example.

New users receive three complete analyses for one cat profile, available within 14 days. No payment card is required. An analysis is counted only after CatMind successfully completes the result.

One credit pays for one successful detailed analysis. Credit packs are a one-time purchase with no monthly billing, and purchased credits do not expire. You can top up when you choose.

Open your Dashboard and select Invite & Earn to copy your private referral link. A friend who registers through that link can earn one additional analysis credit after verifying their email and completing their first successful analysis. Referral attribution is saved during registration, so your friend should begin with the link you shared.

You earn one promotional credit when the invited account either makes its first confirmed purchase or completes three successful analyses across at least two UTC dates. One inviter reward is available for each invited account. Repeated analysis or payment events cannot create duplicate rewards.

Yes. A beneficiary can receive no more than five referral credits in one UTC calendar month. Self-referral, duplicate accounts and coordinated manipulation are prohibited. Shared IP or device hashes are review signals only and never automatically reject a referral. CatMind may review suspicious activity and revoke an unused suspicious award with a recorded compensating ledger entry. See the Terms of Service for the complete programme conditions.

Credits suit occasional use and remain available until used. A subscription provides a larger monthly allowance and a defined number of cat profiles. Monthly analyses reset each billing cycle, and subscribers can add one-off credits when needed.

A technical failure should not consume a trial analysis, monthly allowance or purchased credit. CatMind counts an analysis only when processing completes successfully.

You can wait for the next billing cycle or purchase a one-off credit pack. Daily processing safeguards may also apply independently of your monthly allowance to keep the service responsive for everyone.

Yes. You can cancel at any time. Access normally remains active until the end of the paid billing period. Current prices and allowances are listed on the Pricing page.

CatMind processes account information, details you add to pet profiles, uploaded media, extracted audio and frames, derived analysis data, optional feedback and the technical information needed to operate and secure the service. The exact data depends on the features you use.

CatMind uses measures including password hashing, authenticated account access, access controls, data minimisation, isolated analysis-job storage and deletion workflows. No online system can guarantee complete security, so use a unique password, protect your device and log out on shared devices.

No. CatMind does not sell personal data for money. Information is disclosed only where needed to operate and protect the platform, with your direction or consent, or where required by law. See the Privacy Policy for full details.

Temporary processing files are deleted after the workflow completes or fails, subject to operational cleanup. The current analysis and its media may remain while available in your account, and a later successful analysis can replace earlier full media and results. Protected backups and limited security logs follow their own retention cycles.

When automated inactivity cleanup is enabled, a free-only account may be scheduled for deletion after 90 days of inactivity if its trial has been used and it has no subscription, purchased credits or promotional credits. CatMind plans to provide advance notices approximately 30 and 7 days before deletion. Signing in or choosing to keep the account cancels the schedule. The short inactivity rule does not apply to accounts that still hold purchased credits.

No. Optional dataset contribution requires an explicit choice in the interface. General use of CatMind does not by itself opt an analysis into a training or validation dataset. Eligible exports exclude account details, the pet's name, the original filename and free-text notes as described in the Privacy Policy.

No. CatMind does not use human facial geometry to identify people, authenticate users or create human biometric identity profiles. Incidental people or voices may still be processed as ordinary pixels or sound when they appear in an uploaded recording, so avoid capturing them where possible.

The current platform does not deploy analytics cookies, advertising cookies, remarketing pixels or third-party marketing trackers. It uses first-party browser storage for the selected theme and authentication information after login. The Cookies Policy describes the current implementation.

CatMind provides deletion workflows for analysis history, personal behavioural memory, pet profiles and the user account. Some limited information may remain temporarily in protected backups or where retention is legally required. Privacy requests can also be sent to support@catmind.ai.