AI subprocessors
Model providers, inference platforms and AI tooling that process prompts, files or other customer data. Buyers look for these first on an AI product's subprocessor list, and most ask whether the provider trains on what it receives.
These 86 vendors appear on real companies' published subprocessor lists, with the purpose they were listed for.
Listing one on your own subprocessor list
Give the vendor's legal name, a short purpose such as"Large language model API for in-product AI features", and where it processes your customers' data. Check the vendor's own DPA or subprocessor page for the region you use.
| Vendor | Listed for | Website |
|---|---|---|
 Airon | Use of language models running in Sweden | airon.ai |
 Amazon Bedrock | Power generative AI products | aws.amazon.com |
| Amazon Q | AI-enabled functionality | aws.amazon.com |
 Anthropic | Engineering | www.anthropic.com |
 Arize | AI model monitoring platform | arize.com |
 AssemblyAI | Engineering | www.assemblyai.com |
 Azure OpenAI Service | Large language model service | azure.microsoft.com |
 Berget AI | Use of language models running in Sweden | berget.ai |
 Brain Corp | Process data collected by Physical AI | braincorp.com |
 Bria Artificial Intelligence Limited | AI model provider | bria.ai |
 Browserbase | Application AI Functionality | browserbase.com |
 ByteDance | AI and media processing | bytedance.com |
 Cartesia | Text to speech | cartesia.ai |
 Cassidy | AI prompt orchestration | cassidyai.com |
 Chatbase | AI chatbot hosting | chatbase.co |
 ChatGPT | Large language model API | chatgpt.com |
 Chroma | Vector Inference | trychroma.com |
 Cogito tech | AI training and quality assurance | cogitotech.com |
 Cohere | Engineering | cohere.com |
 Colossyan | AI Video Generation | colossyan.com |
 Corti ApS | AI service | corti.ai |
 Daytona | AI sandbox execution | daytona.io |
 Deep Infra | AI inference | deepinfra.com |
 Deepgram | Speech-to-text services | deepgram.com |
 DeltaGen.AI | AI Support for Technical Questions | deltagen.ai |
 Dust | AI agent platform | dust.tt |
 ElevenLabs | Engineering | elevenlabs.io |
 Fal | AI model inference | fal.ai |
 Fireworks AI | LLM provider/hosting | fireworks.ai |
 Forethought | AI support automation | forethought.ai |
 Gladia | Audio transcription | gladia.io |
 Google Gemini | Large language model API | gemini.google.com |
 Groq | LLM inference | groq.com |
 Helicone | AI Traffic Routing and Debugging | helicone.ai |
 HeyGen | AI video and avatar API | heygen.com |
 Hive | AI Safety | thehive.ai |
 Hugging Face, Inc. | AI model hosting platform | huggingface.co |
 HumanlyHR, Inc. | Conversational AI/Chat-bot services | humanly.io |
 Ideogram | AI inference | ideogram.ai |
 Jasper | AI content generation | jasper.ai |
 Kapa.ai Inc. | AI Search | kapa.ai |
 Labelbox | AI training data platform | labelbox.com |
 Lakera | AI Security | lakera.ai |
 LawY | AI legal research assistant | lawy.ai |
 LlamaIndex | AI & Machine Learning Services | llamaindex.ai |
 Mem0 | LLM Memory Inference | mem0.ai |
 Mentr | AI summary & analysis video meetings | mentr.ai |
 Midjourney | AI Image generation | midjourney.com |
 MiniMax | Text-to-Speech Model Provider | minimax.io |
 Mistral AI | AI inference | mistral.ai |
 OpenAI | Large language model API | openai.com |
 Paradox, Inc. | Conversational AI/Chat-bot services | paradox.ai |
 Parallel Web Systems | AI powered web searches | parallel.ai |
 Perplexity | AI-powered web search and research | perplexity.ai |
 Pipio | AI avatars creation | pipio.ai |
 Raindrop.ai | LLM operations | raindrop.ai |
 Relace | AI model provider | relace.ai |
 Relevance AI | AI agent deployment platform | relevanceai.com |
 Replicate | AI Image Generation | replicate.com |
 Respeecher | Text to Speech | respeecher.com |
 Retell AI | Supports Phone-Based AI Features | retellai.com |
 Rev.com, Inc. | Video transcription for some languages | rev.com |
 Ribbon | AI interviews | ribbon.ai |
 Runway AI, Inc. | AI model provider | runwayml.com |
 Sampark Inc. (Composio) | AI integration platform | composio.dev |
 Serper | LLM operations | serper.dev |
 Shotstack | AI Video Generation | shotstack.io |
 Sieve | AI video processing | sievedata.com |
 Speakeasy Development, Inc. | AI Services | speakeasy.com |
 Speechify | Text to Speech | speechify.com |
 Speechmatics | Transcription | speechmatics.com |
 Stability AI | Generative AI services. | stability.ai |
 Tavus | AI voice agent platform | tavus.io |
 ThoughtSpot | AI analytics and intelligence | thoughtspot.com |
 Together AI | AI infrastructure and model deployment | together.ai |
 Topaz Labs LLC | AI model provider | topazlabs.com |
 Vapi | Audio transcription | vapi.ai |
 VEED | AI models and APIs | veed.io |
 Vellum | AI orchestration platform | vellum.ai |
 Verda | Language model service | verda.com |
 Vertex AI | AI Provider | cloud.google.com |
 Voyage AI | AI embedding and retrieval models | voyageai.com |
 Weights & Biases | AI model training platform | wandb.ai |
 WellSaid Labs | Text to Speech | wellsaid.io |
 xAI | Conversational AI capabilities | x.ai |
 Zenrows | LLM operations | zenrows.com |
Last refreshed 17 July 2026. Each purpose is the one a vendor is usually listed for. Notice Labs doesn't check what a vendor does for any company, so confirm the purpose and location for your own use.
Logos provided by Logo.dev.
What buyers ask about an AI subprocessor
Listing the vendor is the start. These are the follow-up questions a security or privacy reviewer usually sends next.
Does the provider train on our data?
The first thing a buyer asks. Most model APIs sold to businesses say they don't train on API inputs by default, but the terms differ by product and plan. State what your agreement with each provider says, not what the provider's marketing page implies.
How long does the provider keep prompts and outputs?
Many providers keep API traffic for a limited period for abuse monitoring, and some offer zero-retention arrangements on request. Buyers want the period in writing, so check the provider's data processing terms for the one that applies to you.
Where is the data processed?
Region matters for AI vendors as much as for hosting. Some providers let you pin processing to a region, often only on particular plans or through a cloud marketplace. List the region you actually use.
What customer data reaches the model?
Be specific in the purpose column: "summarising support tickets" tells a reviewer far more than "AI features", and lets them judge what personal data could be in a prompt.
An AI trust center puts these answers next to the list: model providers as subprocessors, and controls such as no training on customer data, stated in plain language.