Labs / Time machine / Macroscope

Macroscope

The subprocessor list as we last read it, on Nov 20, 2026.

Dates are when we read the list (UTC), not when Macroscope changed it. Between two readings we don't know what it said.

Sep 30 Nov on Sep 30: first reading Oct 6: document changed, list as published the same Oct 8: document changed, list as published the same Oct 11: read, list unchanged Oct 12: read, list unchanged Oct 13: document changed, list as published the same Oct 14: read, list unchanged Oct 15: read, list unchanged Oct 16: read, list unchanged Oct 17: read, list unchanged Oct 18: read, list unchanged Oct 19: read, list unchanged Oct 20: read, list unchanged Oct 21: read, list unchanged Oct 22: read, list unchanged Oct 23: read, list unchanged Oct 24: read, list unchanged Oct 25: read, list unchanged Oct 26: read, list unchanged Oct 27: read, list unchanged Oct 28: read, list unchanged Oct 29: read, list unchanged Oct 30: read, list unchanged Oct 31: document changed, list as published the same Nov 1: read, list unchanged Nov 2: read, list unchanged Nov 3: read, list unchanged Nov 4: read, list unchanged Nov 5: list changed Nov 6: read, list unchanged Nov 7: read, list unchanged Nov 8: read, list unchanged Nov 9: read, list unchanged Nov 10: read, list unchanged Nov 11: read, list unchanged Nov 12: read, list unchanged Nov 13: read, list unchanged Nov 14: read, list unchanged Nov 15: read, list unchanged Nov 16: read, list unchanged Nov 17: read, list unchanged Nov 18: read, list unchanged Nov 19: read, list unchanged Nov 20: read, list unchanged Sep 30 Nov on Sep 30: first reading Oct 6: document changed, list as published the same Oct 8: document changed, list as published the same Oct 11: read, list unchanged Oct 12: read, list unchanged Oct 13: document changed, list as published the same Oct 14: read, list unchanged Oct 15: read, list unchanged Oct 16: read, list unchanged Oct 17: read, list unchanged Oct 18: read, list unchanged Oct 19: read, list unchanged Oct 20: read, list unchanged Oct 21: read, list unchanged Oct 22: read, list unchanged Oct 23: read, list unchanged Oct 24: read, list unchanged Oct 25: read, list unchanged Oct 26: read, list unchanged Oct 27: read, list unchanged Oct 28: read, list unchanged Oct 29: read, list unchanged Oct 30: read, list unchanged Oct 31: document changed, list as published the same Nov 1: read, list unchanged Nov 2: read, list unchanged Nov 3: read, list unchanged Nov 4: read, list unchanged Nov 5: list changed Nov 6: read, list unchanged Nov 7: read, list unchanged Nov 8: read, list unchanged Nov 9: read, list unchanged Nov 10: read, list unchanged Nov 11: read, list unchanged Nov 12: read, list unchanged Nov 13: read, list unchanged Nov 14: read, list unchanged Nov 15: read, list unchanged Nov 16: read, list unchanged Nov 17: read, list unchanged Nov 18: read, list unchanged Nov 19: read, list unchanged Nov 20: read, list unchanged

tall: list changed · medium: document changed · short: read, no change · x: check failed · dotted: change dates only. Tap a tick or use the date box.

We record change dates, not every check, before Oct 11, 2026.

Read Nov 20, 2026 · 27 companies named as a subprocessor, in 28 entries · stored document

NamePurposeLocationFirst read
Anthropic AI
Large Language Model (LLM) processing to generate summaries and code reviews; Note: None of our customer data is ever used to train any of the Anthropic models. Transfers are safeguarded by Standard Contractual Clauses. Customers have the option to limit data and inference to US only for an extra cost. No data is retained for all models except Fable, where data may be retained for 30 days for security purposes, and this data may be stored globally.
Global
AWS AI
LLM Inference Processing for the product; We occasionally use AWS Bedrock as a backup provider to run inference for other provider LLMs. We also sometimes use AWS for compute services. Customers have the option to limit LLM data and inference to US only for an extra cost.
Global
Baseten AILLM inference and training; We occasionally use Baseten to provide LLM inference and training. We never use customer data to train LLMs.Global
Cloudflare Generating artifacts as part of Murmur; We use Cloudflare to hose generated artifacts from Murmur agents Effective Nov 14Global
Fireworks AI AIInference; We use Fireworks to process inference for certain LLM calls. Models are never trained on any customer data.Global
Google Cloud Platform Cloud provider; Our infrastructure is primarily hosted on GCP.United States
Google Gemini AI
Large Language Model (LLM) processing; We occasionally use Gemini models through Google's Vertex API to generate summaries and do code review. Models are never trained on any customer data.
Global
Linear
Collaboration; We use Linear as our internal project management system. When issues relate to the experience of a specific customer, there may be minimal customer information as part of the created ticket.
United States
Meta For running AI inference of Meta models; We use Meta for inference of their proprietary models. Effective November 9Globally
Microsoft Azure Cloud
Inference and Cloud Services; We will occasionally use Azure to handle inference for LLMs from other providers. We also occasionally use Azure services for cloud processing as a backup to GCP. Customers have the option to limit data and inference to US only for an extra cost.
Global
Modal Compute Sandboxes; We use modal to create sandboxes where we can perform secure computing. We also occasionally us them to process inference for LLMs.Global
OpenAI AI
Large Language Model (LLM) processing to generate summaries and code reviews; Note: None of our customer data is ever used to train any of the OpenAI models. Transfers are safeguarded by Standard Contractual Clauses. Customers have the option to limit data and inference to US only for an extra cost.
Global
Parallel AI powered web searches; Allows Macroscope to browse and search the web.United States
Posthog We use Posthog for analytics and error tracking of our web application and corporate website. Effective date: November 5, 2025; Web analytics and error reportingUnited States and EU
Prime Intellect LLM Inference and Training; We occasionally use Prime Intellect to provide LLM inference and training. We never use customer data to train LLMs. Effective date: August 7, 2026USA
Pylon Customer Support; We use Pylon as our unified customer support platform, to facilitate support via Email, Slack and other mediums.United States
Sail LLM Inference and Training; We occasionally use Sail to provide LLM inference and training. We never use customer data to train LLMs.Global
SendGrid
Email delivery; SendGrid powers email delivery for some of our reporting functionally. This functionally is optional but when configured by customers, allows you to receive summaries via email.
United States
Sentry Error tracking; We use Sentry to track client side errors, which sometimes include minimal customer information to track the source of the issue.United States
Slack Collaboration; We use Slack for internal collaboration, including debugging customer-reported issues which occasionally include customer information.United States
SpaceXAI AILLM Inference and Training; We occasionally use SpaceXAI to provide LLM inference and training. We never use customer data to train LLMs.Global
Stripe Inc Payment processing; Stripe stores customer payment information and processes payments for us.Multi-region
Temporal
Distributed software workflows; Temporal is an integral part of our infrastructure. Any customer data processed via Temporal is encrypted, and Temporal does not have decryption keys.
United States
Terminal We occasionally use Terminal to provide LLM inference and training. We never use customer data to train LLMs. Effective date: September 8, 2026LLM Inference and Training
Together AILLM Inference and Training; We occasionally use Together to provide LLM inference and training. We never use customer data to train LLMs.Global
TypeSafe AI AI Inference; Direct use of AI models for inference. Effective November 12Global
WorkOS In some cases, for Enterprise customers, we leverage WorkOS to integrate with your Identity Provider systems to better improve our summaries.; Integration with Identity ProvidersUnited States
Zapier We use Zapier to automate internal workflows and triage customer support requests.; Workflow automationUnited States

Built from the changes we published; changes we recorded without an alert (parser fixes, entries approved after review) don't appear. Where we later improved how we read a list, an earlier list may look different from the vendor's page at the time.

Gaps in our readings

No gaps since Oct 11, 2026.

All readings (44)
DateResult
read, unchangedviewNov 20, 2026 comparewith Nov 20, 2026
read, unchangedviewNov 19, 2026 comparewith Nov 19, 2026
read, unchangedviewNov 18, 2026 comparewith Nov 18, 2026
read, unchangedviewNov 17, 2026 comparewith Nov 17, 2026
read, unchangedviewNov 16, 2026 comparewith Nov 16, 2026
read, unchangedviewNov 15, 2026 comparewith Nov 15, 2026
read, unchangedviewNov 14, 2026 comparewith Nov 14, 2026
read, unchangedviewNov 13, 2026 comparewith Nov 13, 2026
read, unchangedviewNov 12, 2026 comparewith Nov 12, 2026
read, unchangedviewNov 11, 2026 comparewith Nov 11, 2026
read, unchangedviewNov 10, 2026 comparewith Nov 10, 2026
read, unchangedviewNov 9, 2026 comparewith Nov 9, 2026
read, unchangedviewNov 8, 2026 comparewith Nov 8, 2026
read, unchangedviewNov 7, 2026 comparewith Nov 7, 2026
read, unchangedviewNov 6, 2026 comparewith Nov 6, 2026
read, list changedviewNov 5, 2026 comparewith Nov 5, 2026
read, unchangedviewNov 4, 2026 comparewith Nov 4, 2026
read, unchangedviewNov 3, 2026 comparewith Nov 3, 2026
read, unchangedviewNov 2, 2026 comparewith Nov 2, 2026
read, unchangedviewNov 1, 2026 comparewith Nov 1, 2026
read, document changedviewOct 31, 2026 comparewith Oct 31, 2026
read, unchangedviewOct 30, 2026 comparewith Oct 30, 2026
read, unchangedviewOct 29, 2026 comparewith Oct 29, 2026
read, unchangedviewOct 28, 2026 comparewith Oct 28, 2026
read, unchangedviewOct 27, 2026 comparewith Oct 27, 2026
read, unchangedviewOct 26, 2026 comparewith Oct 26, 2026
read, unchangedviewOct 25, 2026 comparewith Oct 25, 2026
read, unchangedviewOct 24, 2026 comparewith Oct 24, 2026
read, unchangedviewOct 23, 2026 comparewith Oct 23, 2026
read, unchangedviewOct 22, 2026 comparewith Oct 22, 2026
read, unchangedviewOct 21, 2026 comparewith Oct 21, 2026
read, unchangedviewOct 20, 2026 comparewith Oct 20, 2026
read, unchangedviewOct 19, 2026 comparewith Oct 19, 2026
read, unchangedviewOct 18, 2026 comparewith Oct 18, 2026
read, unchangedviewOct 17, 2026 comparewith Oct 17, 2026
read, unchangedviewOct 16, 2026 comparewith Oct 16, 2026
read, unchangedviewOct 15, 2026 comparewith Oct 15, 2026
read, unchangedviewOct 14, 2026 comparewith Oct 14, 2026
read, document changedviewOct 13, 2026 comparewith Oct 13, 2026
read, unchangedviewOct 12, 2026 comparewith Oct 12, 2026
read, unchangedviewOct 11, 2026 comparewith Oct 11, 2026
read, document changedviewOct 8, 2026 comparewith Oct 8, 2026
read, document changedviewOct 6, 2026 comparewith Oct 6, 2026
first readingviewSep 30, 2026 comparewith Sep 30, 2026

History starts Sep 30, 2026. Dates are UTC. Data CC BY 4.0. Shown as stored when this page was built.