These AI training data categories describe the kind of speech, text, image, video and multimodal data we can produce to a brief. Each one maps to one of our eight services and starts from a brief: the language, the domain and the volume. People from our 30,000+ network do the work, and a small sample comes free before you commit.
Browse the full language directory for 100 languages, grouped by region. For speech that's already recorded, see our OTS speech datasets.
An AI training data category is a description of a kind of data a model needs, such as call center speech in Hindi or retail shelf photos, that we produce to a brief rather than sell as a ready-made file.
Most vendors that publish a "datasets" page are showing an inventory: files you can browse and buy. These categories work differently. Every category below is a starting point for a conversation about what your model actually needs, then work that our network of 30,000+ people carries out and we review before delivery.
If you already know the service you need, go straight to our eight services. If you're not sure what data type fits, our pages on speech data, image data, video data and text data cover each type in general.
Everything below applies whether a brief is a single small sample or a multi-language, multi-country rollout.
Alongside the categories we build to a brief, we hold around 35,000 hours of OTS (off-the-shelf) speech, prompts and conversations, roughly 500 hours per language. It's already recorded, every speaker consented to commercial AI training, and sample clips come back in reply to your request.
These 55 categories are examples of what we build across our eight services. Filter by data type, language or domain, or tell us what you need through the brief if you don't see it here.
Example request: A Hindi BFSI call center speech category is a request for recorded banking calls in Hindi: card blocks, loan queries, KYC checks, the kind of speech a bank's voice bot or call center tool has to recognize.
Service Conversational AI data
Example request: English e-commerce customer support text data is a request for the kind of writing a shopping app's support team handles every day: order queries, return requests, delivery complaints, refund chases.
Service Data annotation and labeling
Example request: A retail shelf and product photo category is photography of shelves, aisles and packaging, shot the way a store associate's phone or a fixed shelf camera actually sees them, then boxed and labeled for the products in frame.
Service Computer vision
Example request: An automotive driving scene category is dashcam and road footage: highway, city traffic, parking maneuvers, pedestrians crossing, filmed for a driver assistance or road-scene model rather than pulled from a stock library.
Service Data collection
Example request: A healthcare conversation category is recorded speech between a patient and a provider: appointment calls, symptom checks, follow-up conversations, built for a specific brief and transcribed for accuracy.
Service Transcription
Example request: A multimodal support category pairs the two channels a real support case moves through: a written chat and a follow-up call on the same issue.
Service Conversational AI data
Enterprise email and ticket log data: internal email threads and support ticket logs collected and organized to your brief for AI projects.
Service Data collection
Enterprise form and document data: scanned and digital forms and documents collected and structured to your brief for a specific AI project.
Service Data collection
Enterprise spreadsheet and CRM data: spreadsheet exports and CRM records collected and organized to your brief for a specific AI project.
Service Data collection
Request automotive exterior damage photo data: dents, scratches and panel damage, captured and labeled for detection and classification models.
Service Computer vision
Request automotive interior cabin photo data, captured inside the vehicle and labeled for object and occupancy detection.
Service Computer vision
Request automotive parts and assembly line photo data, labeled for quality inspection and defect detection models.
Service Computer vision
Request healthcare facility and PPE compliance photo data: facility and equipment photos labeled for detection, not diagnostic imaging.
Service Computer vision
Request retail e-commerce catalog photo data, product photos captured and labeled for classification and tagging.
Service Computer vision
Request retail store and checkout photo data, captured in-store and at checkout counters, labeled for detection and classification.
Service Computer vision
Request retail warehouse and inventory photo data: shelf, bin and pallet photos labeled for detection and counting.
Service Computer vision
Automotive dashcam video and voice alert data pairs driving footage with spoken alert audio, collected and reviewed to your brief.
Service Data collection
Image and text product listing data pairs a product photo with its written description, sourced and reviewed to your brief for retail AI.
Service Data collection
Retail shelf photo and inventory text data pairs a shelf image with its stock record, collected and reviewed to your brief for retail AI.
Service Data collection
Hindi chat and voice support data pairs a written chat with a follow-up call in Hindi on the same case, built to your brief for conversational AI.
Service Conversational AI data
Video and transcript captioning data pairs a video with a written transcript and caption track, collected and reviewed to your brief.
Service Data collection
Request Hindi automotive in-cabin voice data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request Japanese BFSI call center speech data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request Spanish BFSI call center speech data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request Tamil BFSI call center speech data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request Arabic customer support call speech data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request German customer support call speech data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request Mandarin customer support call speech data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request Tamil healthcare conversation speech data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request Hindi IVR and voice bot speech data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request Indonesian IVR and voice bot speech data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request Spanish IVR and voice bot speech data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request French retail and e-commerce support speech data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request Portuguese retail and e-commerce support speech data: dialogue recorded, transcribed and reviewed by people, built to your brief, not pulled from stock.
Service Conversational AI data
Request Hindi BFSI chat support text data: banking and finance chat messages written and labeled by people for intent and sentiment.
Service Conversational AI data
Request Tamil BFSI complaint classification data: banking and finance complaints written and labeled by people for category and urgency.
Service Data annotation and labeling
Request German customer support chatbot text data: dialogue turns written and labeled by people for intent and reply quality.
Service Conversational AI data
Request English healthcare patient messaging text data: appointment and scheduling messages written and labeled by people, non-clinical.
Service Data annotation and labeling
Request Spanish chatbot intent classification data: utterances written and labeled by people for intent and entity, built for IVR and bots.
Service Data annotation and labeling
Request Portuguese retail customer feedback text data: feedback written and labeled by people for sentiment and category.
Service Data annotation and labeling
Request French retail review sentiment data: product reviews written and labeled by people across positive, negative and mixed tone.
Service Data annotation and labeling
Request Arabic social media sentiment data: posts and comments written and labeled by people across positive, negative and mixed tone.
Service Data annotation and labeling
Request Mandarin support ticket triage text data: tickets written and labeled by people for category and priority.
Service Data annotation and labeling
Request automotive night driving video data: low-light footage filmed and reviewed to your brief for vision and detection models.
Service Data collection
Request automotive parking and maneuvering video data: low-speed footage filmed and reviewed to your brief for detection and tracking models.
Service Computer vision
Request customer support video call data: video call interactions filmed and reviewed to your brief for quality and evaluation models.
Service Data and model evaluation
Request healthcare telemedicine video data: consultation-style video calls filmed and reviewed to your brief for interaction and quality models.
Service Data collection
Request retail checkout video data: checkout-area footage filmed and reviewed to your brief for object and action detection models.
Service Computer vision
Request retail in-store behavior video data: aisle and store footage filmed and reviewed to your brief for movement and layout analytics.
Service Data collection
Request retail store walkthrough video data: full-store footage filmed and reviewed to your brief for layout and navigation models.
Service Data collection
Egocentric household task video: first-person footage of cooking, cleaning and tidying in real homes, filmed to your brief and labeled for hands, objects and steps.
Service Data collection
Egocentric hand-object interaction video: close-up first-person footage of hands grasping and using objects, filmed to your brief and labeled for hands, objects and contact.
Service Data collection
Egocentric factory and warehouse video: first-person footage of assembly, picking, packing and inspection, filmed to your brief and labeled step by step.
Service Data collection
Egocentric walking and navigation video: first-person footage of streets, buildings and transit, filmed at eye level on smart glasses or head mounts and labeled to your brief.
Service Data collection
Egocentric shopping video: first-person footage of people browsing aisles, picking up products and paying, filmed to your brief and labeled for products and actions.
Service Data collection
The samples on these cards are illustrations, not client data. Every category is available in other languages on request.
Examples across the domains we already work in: BFSI, healthcare, retail and e-commerce, automotive and customer support. Not a full list, and not a claim that these are the only combinations we handle.
An example combination across a data type and a domain we already cover.
An example combination across a data type and a domain we already cover.
An example combination across a data type and a domain we already cover.
An example combination across a data type and a domain we already cover.
An example combination across a data type and a domain we already cover.
An example combination across a data type and a domain we already cover.
An example combination across a data type and a domain we already cover.
An example combination across a data type and a domain we already cover.
More general terms are covered in our AI training data glossary.
A category is a description of a kind of data we can produce for a model, such as Hindi call center speech or retail shelf photos. It's a starting point for a brief, not a dataset sitting ready to ship.
No. Nothing on this page is a stock file. Every category is produced to your brief: the language, domain and volume you need.
Yes. The 55 categories here are examples across our eight services. Tell us what you need through the brief and we'll scope it.
We work in languages worldwide, with deep coverage of Indian languages beyond Hindi and English. Pick a language chip below to see which categories use it as their lead example, or browse our full language directory; every category can be adapted to other languages on request.
Each category maps to one of our eight services, shown on its card and its own page. Data collection, annotation, transcription, conversational AI data, evaluation, computer vision, translation and synthetic data all cover different categories.
Yes. Every category page has a brief you can send, and we produce a small sample free before you commit to the full scope.
Pick the category or describe your own, add the languages and a rough volume, and send it. We'll produce a small sample free before you commit.
This opens your email app with the brief filled in. Nothing is sent until you press send. You can also use the contact page.