Dia
Dia uses 20 subprocessors and infrastructure vendors, including AWS, Google Cloud and OpenAI, based on its public subprocessor list. Last checked Nov 16, 2026.
AI Stack
Foundation Models
AI assistant processing: Anthropic's Claude models power AI features in Dia, including chat and tab-aware assistance. When members use these features, we send the relevant prompts and context through Atlassian's AI Gateway to Claude so it can generate responses. Claude requests are usually processed on Amazon Bedrock, and may be routed to Google Vertex AI or to Anthropic directly when needed for reliability. Anthropic does not train on Dia member data and does not retain it beyond what's needed to process each request.
Cloud provider & AI processing: GCP hosts core Dia services and databases, including analytics systems like BigQuery. Member data stored in these systems is used to operate Dia, power features, and run privacy-respecting analytics. We apply strict access controls, encryption, and retention policies. We also use Google Vertex AI to run some AI models, including Gemini and, when needed, Claude, and Google Model Armor for content safety checks. Vertex AI and Model Armor do not train on Dia member data and do not retain prompts or responses beyond what's needed to process each request. Dia also uses Google's search suggestion and translation services. When members use these features, Dia sends their search queries or the text they want translated to Google.
AI assistant processing: OpenAI powers some AI features in Dia. When members use these features, we send the relevant prompts and context to OpenAI, mostly through Atlassian's AI Gateway. Some requests, such as content safety checks, go to OpenAI directly so it can review them. OpenAI does not train on Dia member data and does not retain it beyond what's needed to process each request.
Infrastructure
Cloud & Networking
Cloud provider & AI processing: AWS is where we run much of Dia's backend infrastructure. We use AWS to host and store production systems that power the browser, including services that process and store Dia member data as part of core product functionality. We also use Amazon Bedrock to run AI models, including Anthropic's Claude, which is the primary way Dia accesses Claude. Amazon Bedrock does not train on Dia member data and does not retain prompts or responses beyond what's needed to process each request.
CDN & traffic routing: Cloudflare sits in front of Dia to keep traffic fast and reliable. It routes and optimizes web requests to our services and helps protect against common web attacks. To do this, Cloudflare processes IP addresses, request metadata, and the contents of requests as they pass through its network.
Cloud provider & AI processing: GCP hosts core Dia services and databases, including analytics systems like BigQuery. Member data stored in these systems is used to operate Dia, power features, and run privacy-respecting analytics. We apply strict access controls, encryption, and retention policies. We also use Google Vertex AI to run some AI models, including Gemini and, when needed, Claude, and Google Model Armor for content safety checks. Vertex AI and Model Armor do not train on Dia member data and do not retain prompts or responses beyond what's needed to process each request. Dia also uses Google's search suggestion and translation services. When members use these features, Dia sends their search queries or the text they want translated to Google.
Frontend hosting & serverless infrastructure: Vercel hosts diabrowser.com and other Dia web experiences. When members interact with these components, Vercel processes standard web request data (like IP addresses and interaction metadata) so it can serve pages quickly and reliably.
Observability
Infrastructure monitoring & logging: Datadog helps us monitor the health and performance of Dia. We send infrastructure logs, metrics, and application telemetry so we can detect incidents, troubleshoot issues, and keep the service reliable. These logs may include limited user-related metadata (such as identifiers or IPs), but are used for operations and security and not for advertising.
Crash & error reporting: Sentry collects crash and error reports so we can find and fix bugs in Dia. Reports include technical details like device and app version, stack traces, and limited identifiers.
Analytics & BI
Product analytics: Amplitude helps us understand how people use Dia so we can improve the product. We send product usage events (for example, which features are used and how often), along with user and device identifiers, so we can measure adoption, debug confusing flows, and make better product decisions. We do not sell this data to third parties.
Data analysis: Mode is the tool our team uses to query and analyze data from our analytics warehouse. Query results, which may include product usage data and member identifiers, are stored in Mode so our team can build reports.
Product analytics: Segment is our event collection and routing layer for product analytics. Dia usage events, along with user and device identifiers, are sent to Segment first, then forwarded to tools like Amplitude for reporting. This lets us centralize controls over what data we collect, where we send it, and how long we keep it.
CRM & Support
AI customer support: Sierra powers our AI support assistant. When members contact support, Sierra processes their messages and contact details to answer questions and pass requests to our support team.
Customer support: Zendesk is our support ticketing system. When members contact support, Zendesk stores the messages, contact details, and any attachments they send so our team can respond.
Communications
Transactional emails: SendGrid sends Dia's account and service emails, such as messages about members' accounts and important service notices. To deliver them, SendGrid processes members' email addresses and the contents of each message. We use SendGrid only for emails needed to provide the service. Marketing emails go through Mailchimp.
Marketing emails: Mailchimp sends Dia's marketing and product update emails. When members create a Dia account, we add their email address to our Mailchimp audience, along with tags such as which product they use and their plan or trial status, so we can send relevant emails. Members can unsubscribe at any time. Mailchimp does not receive browsing data or content from Dia.
Developer Platform
Identity & AI routing: Atlassian, The Browser Company's parent company, provides account identity for Dia. New Dia accounts are Atlassian accounts, and members at organizations that manage their accounts through Atlassian sign in through their organization's identity provider. When members sign in, Dia receives their name, email address, and Atlassian account ID, and automatically signs them in to their Atlassian products. Atlassian also routes Dia's AI requests to model providers through its AI Gateway, so prompts and page context that members send to Dia's AI features pass through Atlassian's systems. The AI Gateway does not store the content of prompts or responses. It keeps request logs, such as user ID, organization, model, and token usage, to operate the service.
Feature flag management: LaunchDarkly helps us roll out new features gradually and turn them on or off safely. To decide which features a member sees, Dia sends LaunchDarkly the member's user ID and their app platform and version. LaunchDarkly does not receive names, email addresses, browsing data, or content from Dia.
Internal tools: Retool hosts internal tools our team uses to support members, such as reviewing product feedback and handling account deletion requests. When authorized employees use these tools, Retool processes the member data they display, such as account identifiers and feedback content.
Model labs reached through vendors 2
Dia does not list these model providers. Vendors on its list name them on their own subprocessor lists.
- SpaceXAIthroughCloudflare,Vercel
- MistralthroughZendesk
Changes
Checked Nov 16, 2026Source
This is the first saved copy of Dia’s subprocessor list. Added and removed vendors will show here after the next check.
Similar companies
Companies with the most overlapping vendors