Snapshot 310535

Document
Subprocessor list
URL
https://www.gleap.ai/trust/subprocessors
Fetched
HTTP status
200
Content type
text/html
Fetch mode
static
Size
177505 bytes
SHA-256 (raw)
faba38f79e9a1d0dffee1f235c1987a03ecbc31015a125819361ba4210c7b9c7
SHA-256 (normalized text)
92a1394eaffc406932697176ba7d1f5b607caa9703164912d9298a22e228c932

Raw (normalized text)

Scripts and page chrome removed; this is what change detection compares.

Gleap uses the following sub-processors to provide its services. The contractually authoritative list is Annex 8.10 of our Data Processing Addendum.
Under the revised DPA, customers grant general written authorization for the listed sub-processors. Gleap provides at least 30 calendar days’ advance email notice before an additional or replacement sub-processor engaged directly by Gleap begins processing customer personal data. Customers may object during that period on reasonable data protection grounds. Publication of an updated list does not replace email notice.
For changes to further processors engaged by a sub-processor, Gleap promptly forwards the upstream notice by email, including the applicable notice and objection periods. These periods may be shorter than 30 days. Customers receive a reasonable opportunity to object. If the upstream timetable does not allow that opportunity before the change, or a timely objection remains unresolved when the change takes effect, Gleap suspends affected processing or provides an agreed alternative that avoids the objected-to provider, in accordance with clause 8.4.
Existing agreements that require individual written consent retain that requirement until validly amended.
The provider list was updated on November 18, 2026. Listing a provider does not establish that every provider is used for every customer or remove the DPA’s requirements for international-transfer safeguards. AI model requests from both regions go through OpenRouter, which forwards each request to a provider that hosts the model; the one exception is OpenAI’s speech service for Kai Voice, which Gleap contacts directly. Provider processing is governed by the applicable provider agreements and transfer safeguards.
Regional hosting and provider processing
Gleap supports EU and US data residency. Customer databases and their backups stay in the region chosen at signup. Uploaded files are stored with Cloudflare in separate storage for each region. A provider’s corporate address is not a guarantee of where it processes data. Regional hosting does not automatically localize email, monitoring, billing, integrations or AI processing. The provider-specific locations below apply to those services; see our data residency guide. Existing contractual authorization, notice and transfer-safeguard requirements continue to apply.
Infrastructure & hosting
DigitalOcean, LLC (USA), regional hosting locations: Frankfurt, Germany (EU) and NYC3, New York (US), according to the selected region. Our main cloud provider: application servers, screenshot rendering, the production MongoDB and Valkey databases through DigitalOcean Managed Databases, and Gleap’s search, vector and analytics databases.
Cloudflare, Inc. (USA), server locations: worldwide. File storage for screenshots and attachments uploaded through the Gleap widget and dashboard, kept separately for each region. Global delivery and video-call infrastructure may involve processing outside the selected region.
Communication & delivery
Postmark (AC PM, LLC, USA), server location: USA. Transactional email delivery.
Firebase (Google Ireland Limited, Ireland), push notification delivery. This feature must be manually activated.
Telephony
Twilio Ireland Limited (Ireland), server locations: Ireland (Dublin) for calls of EU accounts and USA for calls of US accounts. Phone numbers and calls for the optional Kai Voice phone feature: call routing, the audio between callers, Kai and teammates, forwarding to numbers the customer sets up, one-time codes by SMS that verify callers, and the transcript of the part of a call with a teammate or a forwarded number (Twilio’s speech engine for it: Google Speech-to-Text; this processing may take place in the USA). Phone number administration, the registration details a number requires and SMS codes are processed in the USA. Calls are not recorded. This feature must be manually activated. Other provider processing and onward transfers are governed by the applicable provider agreement and transfer safeguards.
Billing
Stripe Payments Europe, Limited / Stripe Technology Europe, Limited (Ireland), potential future payment provider for customer billing; not currently used. Any activation remains subject to the applicable contractual authorization and notification requirements.
Paddle.com Market Ltd. (UK), payment provider for customer billing.
Monitoring
New Relic, Inc. (USA), server location: EU (Frankfurt, Germany). Application performance monitoring and logging for Gleap’s infrastructure; telemetry may include request metadata.
Agent integrations
Sampark Inc. d/b/a Composio (USA), integration platform powering custom-agent tool connections (e.g., third-party app actions and triggers configured by the customer).
AI providers (optional)
The following providers support optional AI features, including automated processing associated with enabled features. Administrators can disable individual AI features. AI availability is also subject to the account’s credit balance or legacy AI usage setting. Customers can choose from the models offered for configurable AI features, and Enterprise plans support bring-your-own-model arrangements. Selecting a model does not by itself guarantee a processing region; see the provider details below. Contact us for help configuring AI restrictions for your account. Under our DPA, Gleap shall not use customer personal data to train AI models or authorize providers to use it for that purpose.
OpenRouter, Inc. (USA), gateway for the AI model requests of both regions. Gleap sends its requests for Claude, GPT, Gemini, Grok, Kimi, GLM, DeepSeek, Mistral and other models to OpenRouter, which forwards each request to a provider that hosts the model. Requests may be processed in the EU, United States, or China. Selecting a model does not guarantee the downstream hosting provider or processing region.
Google Vertex AI and Amazon Bedrock, reached through OpenRouter. Hosting of Claude models: Claude requests go to Google Vertex AI first, then to Amazon Bedrock.
Microsoft Azure, reached through OpenRouter. Hosting of GPT models: GPT requests of EU accounts go to Azure’s EU region first, then to Azure in other regions.
Model hosts for EU only, reached through OpenRouter’s EU endpoint, all processing in Europe (today all of these hosts process in the EU): Microsoft Azure (EU regions), Google Vertex AI (EU regions), Amazon Bedrock (Ireland), Mistral AI (France) and Inceptron (Sweden, data center in Finland). Through this partnership, organizations that choose EU only can use models from OpenAI, Anthropic, Google, Mistral and open-model families with European inference and zero data retention.
Model hosts for US only, reached through OpenRouter’s US endpoint, all processing in the United States: Microsoft Azure, Google Vertex AI, Amazon Bedrock, DeepInfra, Fireworks, Baseten, Inference.net, Morph, Wafer, Sail Research, Particle AI, xAI and Mistral AI, all with zero data retention, and Alibaba Cloud (US region) only for the Qwen models that no host offers with zero data retention.
Other model hosts (Global), reached through OpenRouter. For organizations on Global, OpenRouter chooses among the hosts on its zero data retention list, which may process requests in the EU, the United States or China.
The model hosts above are the ones OpenRouter offered for these regions on November 18, 2026. OpenRouter adds models and hosts over time; Gleap forwards such changes as described in our DPA. For the current models and the hosts that serve each one, see OpenRouter’s models and providers.
OpenAI Ireland Ltd (Ireland), contacted directly only for speech, through OpenAI’s default API endpoint for both regions: the real-time voice sessions of the optional Kai Voice feature (speech recognition and speech generation in the widget and on phone calls) and generated speech such as phone greetings; processing may take place outside the selected region. GPT models for other features are requested through OpenRouter.
Training and retention. Gleap’s model requests instruct OpenRouter to use only hosting providers that do not train on the request data. Zero data retention is requested for most requests, not for all. It is not requested for models that no provider offers with it (currently Claude Fable 5.1, Claude Fable 5, Qwen 3.6 Plus and Qwen 3.8 Flash), for embedding and reranking requests, or for speech-to-text and image generation, and it is not guaranteed for the code analysis service and coding sandboxes of Kai Code. The four models named above are not available with zero data retention and are marked in the model picker; their requests still go only to providers that do not train on the data, and every other model in the picker is requested with zero data retention. The speech processing of Kai Voice at OpenAI does not go through OpenRouter and is not covered by these instructions.
AI processing region and EU endpoints. Each organization chooses where its AI agents run: EU only or Global for EU accounts, US only or Global for US accounts. The choice is offered at signup and can be changed in the organization’s privacy settings. With EU only, Gleap sends the agents’ requests to OpenRouter’s EU endpoint, which keeps prompts and completions in Europe (the EU/EEA, the UK and Switzerland) and returns an error rather than process a request outside it (OpenRouter In-Region Routing). This gives EU accounts European inference with zero data retention: every model in the picker is requested with zero data retention, except the four models named above, which still run in the EU without it. With US only, the requests go to OpenRouter’s US endpoint in the same way. If the region does not serve the selected model, a model the region serves is used instead. The setting covers Kai and the AI agents (support, copilot, custom and coding agents, AI tasks and the matching of duplicate feature requests). Other AI features, such as summaries, tagging, translation, embeddings, reranking and media processing, are not bound to this setting, and neither is the speech processing of Kai Voice at OpenAI.
AI data processing
These services support Gleap’s AI features (knowledge base, Kai agents, and Kai Code) and process data only when those features are used:
MongoDB, Inc. (Voyage AI) (USA), Voyage AI embedding and reranking models for knowledge-base content, tickets, and messages used in AI search and retrieval, requested through OpenRouter. They are used for AI inference, not production database hosting.
FoundryLabs, Inc. (E2B) (USA), isolated cloud sandboxes in which Kai Code works on repositories the customer has connected.
Context Dev Inc. (Context.dev) (USA), used only to fetch public websites: crawling a customer’s public website for knowledge-base ingestion, reading public web pages for Kai agents, and looking up a website’s public logo and colors during setup.
TinyFish, Inc. (USA), web search and page retrieval for Kai agents.
Endless Labs, Inc. (Datalab) (USA), conversion of uploaded documents (PDFs, spreadsheets) to machine-readable text for AI processing.
Questions
For questions about this list, data locations, or to be notified of sub-processor changes, email privacy@gleap.ai.