Subprocessors

Last updated: October 30, 2026

Overview

This page identifies service providers and customer-directed integration providers that may process Customer Content or personal information in connection with Capy. It supplements our Privacy Policy, Terms of Service, and Data Processing Addendum.

Not every provider is used for every customer, plan, feature, model, integration, or region. Data processed depends on your configuration, selected models, connected repositories, enabled integrations, support requests, billing setup, and instructions to agents.

Core subprocessors

These providers may process personal information or Customer Content on Capy’s behalf to operate the Services.

ProviderPurposeExample dataProcessing location
Amazon Web ServicesCloud infrastructure, storage, encryption, networking, model inference through Amazon Bedrock, and related servicesCustomer Content, files, logs, artifacts, operational metadataUnited States and other AWS regions as configured
CloudflareNetwork, security, edge, DNS, and web-deployment routing servicesIP address, request metadata, deployment routing metadataGlobal
VercelWebsite and web application hosting, edge delivery, preview/deployment infrastructure, analyticsIP address, browser/device data, request metadata, page/app usageGlobal
PlanetScaleProduction database infrastructureAccount, organization, project, task, billing, integration, and application dataUnited States and other regions as configured
ClickHouseTelemetry, product analytics, usage analytics, operational reporting, and agent cloud machine traces and logsUsage events, operational metadata, product telemetry, machine service logsUnited States and other regions as configured
UpstashRedis-compatible caching, queues, rate limits, realtime state, and operational coordinationSession metadata, cache entries, queue metadata, operational dataUnited States and other regions as configured
HatchetWorkflow orchestration and background job executionWorkflow metadata, task state, queued event data, operational metadataUnited States and other regions as configured
Latitude.shBare-metal hosting and object storage for agent cloud machinesRepository checkouts, files, commands, logs, environment configuration, machine snapshotsUnited States
BackblazeArchive storage for dormant machine snapshotsMachine snapshots, including repository checkouts and filesUnited States
turbopufferSearch indexing for threads, pull requests, and issuesMessage text, thread and task titles, pull request and issue textUnited States
RevylCloud phones, mobile app builds, and device sessionsRepository source archives, app builds, device sessions, organization name, owner emailUnited States and other regions as configured
Grafana LabsInfrastructure metrics for agent cloud machinesMachine and team identifiers, host metricsUnited States
ClerkAuthentication, account management, sessions, and organization identityNames, emails, user IDs, session data, organization membershipUnited States and other regions as configured
StripePayment processing, invoices, subscriptions, taxes, fraud prevention, and disputesBilling contact, payment metadata, invoices, transaction IDs, tax dataGlobal
AutumnSubscription management, entitlements, credits, usage-based billing, and billing stateOrganization IDs, plan, credits, usage records, subscription metadataUnited States and other regions as configured
Customer.ioTransactional email, lifecycle email, customer messaging, and email engagementNames, emails, organization data, email events, message metadataUnited States
PylonCustomer support, security reports, and issue-management workflowsSupport messages, contact information, issue metadata, security-report contextUnited States and other regions as configured
AshbyCareers pages, recruiting workflows, job applications, and referral handlingCandidate names, emails, resumes, application materials, referral dataUnited States
SlackInternal alerting and recruiting referral notificationsReferral form data, operational alerts, support or security notificationsUnited States and other Slack regions

AI and model providers

Capy routes prompts, code context, files, logs, images, screenshots, tool output, and generated Output to model providers only as needed to provide the selected AI features. Model availability and routing can change over time.

ProviderPurposeExample dataNotes
AnthropicLanguage model inferencePrompts, code context, files, tool output, generated OutputUsed for Claude models
OpenAILanguage model inference and external subscription featuresPrompts, code context, files, tool output, generated OutputUsed for GPT/Codex models and related features
OpenRouterModel routing and custom model-provider accessPrompts, code context, files, tool output, generated OutputUsed for selected routed and customer-configured models
Vercel AI GatewayModel routing with zero data retentionPrompts, code context, files, tool output, generated OutputUsed for selected routed open models
Google / Google Cloud / Vertex AILanguage model inference and model hostingPrompts, code context, files, tool output, generated OutputUsed for Gemini and Vertex-hosted models
xAILanguage model inferencePrompts, code context, files, tool output, generated OutputUsed for Grok models where available
Fireworks AIModel hosting and inferencePrompts, code context, files, tool output, generated OutputUsed for selected open and hosted models
Z.AILanguage model inferencePrompts, code context, files, tool output, generated OutputUsed for selected GLM models where available
DeepSeek model hostsLanguage model inference where routed through supported providersPrompts, code context, files, tool output, generated OutputUsed for selected DeepSeek models where available
CerebrasModel hosting and inferencePrompts, code context, files, tool output, generated OutputUsed for selected open models where available
InceptionLanguage model inferencePrompts, code context, files, tool output, generated OutputUsed for Mercury models
Moonshot AILanguage model inferencePrompts, code context, files, tool output, generated OutputUsed for Kimi models, routed through OpenRouter or Vercel AI Gateway
Alibaba CloudLanguage model inference and model hostingPrompts, code context, files, tool output, generated OutputUsed for Qwen and selected DeepSeek models, routed through OpenRouter or Vercel AI Gateway
TencentLanguage model inferencePrompts, code context, files, tool output, generated OutputUsed for Hunyuan models, routed through OpenRouter
MetaLanguage model inferencePrompts, code context, files, tool output, generated OutputUsed for Muse Spark models, routed through Vercel AI Gateway
ParticleLanguage model inferencePrompts, code context, files, tool output, generated OutputUsed for Particle models, routed through Vercel AI Gateway
DeepInfraModel hosting and inferencePrompts, code context, files, tool output, generated OutputUsed for selected open models, routed through OpenRouter
Atlas CloudModel hosting and inferencePrompts, code context, files, tool output, generated OutputUsed for selected open models, routed through OpenRouter
Novita AIModel hosting and inferencePrompts, code context, files, tool output, generated OutputUsed for selected open models, routed through OpenRouter
ExaWeb search and retrieval featuresSearch queries, retrieved snippets, task context needed for searchUsed when web search/research tools are available
FirecrawlWeb search and retrieval featuresSearch queries, URLs, retrieved page contentUsed alongside Exa when web search/research tools are available
ElevenLabsSpeech-to-text transcriptionAudio recordings, transcriptsUsed for composer dictation and for transcripts of attached voice messages
fal.aiMedia or image generation features where availableImage prompts, reference images, generated mediaUsed only for media-generation features where enabled

Capy does not use non-public Customer Content to train models that Capy develops or controls unless you explicitly opt in or otherwise agree in writing. Third-party model-provider processing is subject to the applicable provider’s terms, policies, configurations, and data-processing commitments.

When you use BYOK, external subscriptions, customer-provided model credentials, or customer-directed model providers, that provider may process data under your own agreement with that provider rather than as Capy’s subprocessor.

Customer-directed integrations

These providers process data when you connect or instruct Capy to interact with them. They may act as independent controllers, processors for you, or service providers under their own terms.

ProviderPurposeExample data
GitHubRepository access, branch creation, commits, PRs, issues, comments, checks, reviews, and webhooksRepository contents, commit metadata, PR/issue data, comments, user identifiers, installation metadata
SlackSlack-originated tasks, thread sync, message search/read tools, notifications, and file attachmentsWorkspace metadata, channel/thread messages, files, user IDs, permissions, reactions
LinearIssue delegation, context sync, team mappings, project routing, and task updatesIssue text, comments, team/project metadata, user identifiers
SentryError/issue context, automation triggers, and issue investigationIssue metadata, stack traces, event data, project metadata
VercelPreview deployment detection and deployment contextDeployment URLs, PR comments, project/deployment metadata
GitHub Copilot, ChatGPT/Codex, or SuperGrok external subscriptionsCustomer-directed model access using the customer’s own subscriptionPrompts, code context, generated Output, subscription status
MCP servers and custom toolsCustomer-directed tool access and context retrievalData exposed by the configured server or tool
Cal.comSales/demo schedulingName, email, company, scheduling metadata
YC Bookface, Solo Founders, and partner portalsPartner credit eligibility and redemption flowsEmail, company, eligibility metadata submitted through partner workflows

You control whether these integrations are enabled and what permissions are granted. Review each provider’s permissions, privacy policy, and terms before connecting it to Capy.

Recruiting and marketing providers

Our marketing site and careers workflows may use additional providers for job listings, applications, referrals, scheduling, analytics, and communications.

ProviderPurposeExample data
AshbyJob listings, application forms, application submission, recruiting workflowsCandidate data, resumes, applications
Cal.comSales/demo schedulingName, email, company, meeting metadata
PromptwatchMarketing-site visitor analytics, including AI search referral trafficPageviews, referrer, location, device and browser data
Customer.ioEmail delivery and engagementEmail address, message metadata, opens/clicks
ResendCRM email campaigns and reply handlingEmail address, name, message content, replies
Context.devCRM company enrichment from email domainsEmail domains, company details
ExaCRM research on business contacts and companiesNames, email addresses, company details
XSocial outreach from our CRMX handles, direct messages, engagement data
TypeSafeCRM data enrichment and scoring of business contacts and companiesNames, job titles, company details, public profile data
UnipileLinkedIn messages and connection requests for sales outreachLinkedIn profile identifiers, messages, connection status
SlackInternal referral and operational notificationsReferral form content, alert metadata

How updates work

We may update this page at any time in our sole discretion, including when providers are added, removed, replaced, renamed, or used for new purposes. Each change is listed under Changes below and in the subprocessor RSS feed, so you can subscribe to be notified. If a new material Subprocessor is added for Customer Personal Data under the Data Processing Addendum, customers may object as described in that DPA or their applicable agreement.

Questions or objections may be sent to support@capy.ai with the subject line "Subprocessors."

Changes

  • October 30, 2026: Added speech, search, mobile, monitoring, and model providers. ElevenLabs transcribes composer dictation and attached voice messages. Firecrawl serves web search and retrieval alongside Exa. turbopuffer indexes threads, pull requests, and issues for search. Revyl runs cloud phones and mobile app builds. Grafana Labs receives infrastructure metrics for agent cloud machines, and ClickHouse also stores their traces and logs. Added Vercel AI Gateway, Moonshot AI, Alibaba Cloud, Tencent, Meta, Particle, DeepInfra, Atlas Cloud, and Novita AI as model providers.
  • October 30, 2026: Added Resend, Context.dev, Exa, and X as marketing providers. Resend sends CRM email campaigns and handles replies. Context.dev enriches companies from email domains. Exa researches business contacts and companies. X carries social outreach from our CRM.
  • October 15, 2026: Updated infrastructure and model providers. Hatchet replaces Temporal for workflow orchestration. Latitude.sh replaces Daytona as the host for agent cloud machines, and Backblaze stores archived machine snapshots. Added Cerebras and Inception as model providers, and Amazon Bedrock model inference under Amazon Web Services.
  • October 15, 2026: Added TypeSafe and Unipile as marketing providers. TypeSafe enriches and scores business contacts and companies in our CRM. Unipile sends LinkedIn messages and connection requests for sales outreach. Neither processes Customer Content.
  • October 15, 2026: Removed PostHog and Sentry as app subprocessors. The Capy app no longer sends data to PostHog or Sentry. PostHog remains a marketing-site analytics provider, and Sentry remains a customer-directed integration.