# Context > Context is a platform for building, running, and improving AI agents on > infrastructure the customer controls. Workspace is where people define and > review work. Engine runs agents with tools, identity, and isolated compute. > Unify stores the procedures and context those agents use. Evals measures > completed work against criteria set by the customer's team. Key facts: - Deployment: managed, customer VPC (AWS, Azure, or GCP), on-premises, or air-gapped environments designed for disconnected operation. - Models: model-agnostic. Claude, GPT, Gemini, Kimi, or open weights; teams can switch models mid-task without losing files, tools, or prior work. - Identity and audit: every agent is its own member of the organization and works on behalf of the person who asked; people sign in with SAML or OIDC and are provisioned with SCIM. Each run gets its own short-lived key, and every action is recorded in the audit log. Credentials are brokered to connectors at runtime, never written into prompts. - Quality: built-in evals. Expert-authored rubrics score every run; golden sets guard against regressions; accepted work can distill into cheaper models the customer owns. - Deliverables: real, editable files (PowerPoint .pptx presentations, Word documents, research reports) saved with sources and changes attached. - Data: one customer's traces, corrections, and institutional context are never used to train models for other customers. - Pricing: quote-based, shaped by deployment model, usage, and support requirements. Contact: https://www.context.ai/contact ## Products - [Context Code](https://www.context.ai/cli): the Context agent CLI, with terminal sessions that land in the workspace task list - [Context Desktop](https://www.context.ai/desktop): the Context app for Mac, Windows and Linux, with workspaces and your own deployments in tabs and Cascade dictation into any app (download at https://www.context.ai/desktop#download) - [Context Mobile](https://www.context.ai/mobile): the Context workspace on iPhone, with notifications you can act on, a Live Activity and widgets (private preview) - [Context Workspace](https://www.context.ai/product/workspace): Where your team and agents work - [Context Engine](https://www.context.ai/product/engine): The runtime for your agents - [Context Unify](https://www.context.ai/product/unify): Connections for people and agents - [Context Evals](https://www.context.ai/product/evals): Production evals for agent workflows - [Context Drive](https://www.context.ai/product/drive): Cloud storage in Context Office - [Context Docs](https://www.context.ai/product/docs): Documents in Context Office - [Context Sheets](https://www.context.ai/product/sheets): Spreadsheets in Context Office - [Context Slides](https://www.context.ai/product/slides): Presentations in Context Office - [Context PDFs](https://www.context.ai/product/pdf): PDFs in Context Office - [Context Search](https://www.context.ai/product/search): Search in Context Office - [Context Wiki](https://www.context.ai/product/wiki): Team standards and pages every agent follows - [Together](https://www.context.ai/product/together): Team chat where agents are members - [Context in Slack](https://www.context.ai/product/slack): Your workspace's agents in Slack - [Context Teams](https://www.context.ai/product/teams): Agents as members of your team - [Context Skills](https://www.context.ai/product/skills): Skill management for enterprises - [Apps](https://www.context.ai/product/apps): Internal tools built from a request - [Context RL](https://www.context.ai/product/rl): Better skills, harness and model from graded runs - [Secure VM](https://www.context.ai/product/secure-vm): A sandbox per task, inside your perimeter - [Agents](https://www.context.ai/product/agents): AI teammates for your organization - [Tasks](https://www.context.ai/product/tasks): Work you hand to agents - [Projects](https://www.context.ai/product/projects): Folders for your agent work - [Automations](https://www.context.ai/product/automations): Recurring work agents run on a schedule - [Approvals](https://www.context.ai/product/approvals): The decision before an agent acts - [Inbox](https://www.context.ai/product/inbox): What happened while you were away - [Agent Identity](https://www.context.ai/product/identity): An identity for every agent - [Connectors](https://www.context.ai/product/connectors): Your apps and services, connected to agents - [API and MCP](https://www.context.ai/product/api): Context in your code and your AI tools - [Models](https://www.context.ai/product/models): Every leading model, run where you choose - [Admin](https://www.context.ai/product/admin): People, agents and access, managed in one place - [Audit log](https://www.context.ai/product/audit): Every agent action, with who asked and who decided - [Usage and billing](https://www.context.ai/product/usage): What your agents spend, and on what - [Microsoft Teams](https://www.context.ai/product/microsoft-teams): Teams and Microsoft 365, in reach of your agents ## Comparisons Honest, dated, vendor-sourced comparisons with neighboring products: - [All comparisons](https://www.context.ai/compare) - [Context vs ChatGPT Enterprise](https://www.context.ai/compare/chatgpt-enterprise): Context runs agents that do production work inside your security perimeter, on any model, with every run scored and audited. ChatGPT Enterprise gives every employee a hosted assistant. - [Context vs Claude](https://www.context.ai/compare/claude): Context is where your team runs Claude alongside every other major model, with shared workflows, audit, and evals on infrastructure you choose. - [Context vs Microsoft 365 Copilot](https://www.context.ai/compare/microsoft-copilot): Context runs agents across every system your company uses, on infrastructure you control. Copilot brings AI inside the Microsoft 365 tenant. - [Context vs Glean](https://www.context.ai/compare/glean): Context puts what your company knows to work in governed agent workflows, scored on every run and operated inside your perimeter. - [Context vs Zapier Agents](https://www.context.ai/compare/zapier-agents): Context gives each agent its own computer, durable runs, and measured quality for work that used to take a person. Zapier connects the widest app catalog for lightweight automation. - [Context vs Claude Code](https://www.context.ai/compare/claude-code): Context Code runs every model in your catalog, Claude included, and puts every session in your team's task list on one bill. - [Context vs OpenAI Codex](https://www.context.ai/compare/codex): Context Code runs any model in your catalog, reads the AGENTS.md your repositories already have, and reports every session to your workspace. ## Solutions - [Semiconductors](https://www.context.ai/solutions/semiconductors): Turnkey agents for chip design, fab operations, and the supply chain behind every wafer. - [Financial Services](https://www.context.ai/solutions/financial-services): Turnkey agents for the deal team, the desk, and the back office. - [Consulting](https://www.context.ai/solutions/consulting): Turnkey agents for research, modeling, and client-ready deliverables. - [Telecom](https://www.context.ai/solutions/telecom): Turnkey agents for the network, the NOC, and the customer. - [Public Sector](https://www.context.ai/solutions/public-sector): Turnkey agents for programs, records, and citizen services. - [Industrials](https://www.context.ai/solutions/industrials): Turnkey agents for the plant floor, maintenance, and the supply chain. - [Business Operations](https://www.context.ai/solutions/business-operations): Turnkey agents for the work that keeps every team running. - [Insurance BPO](https://www.context.ai/solutions/insurance-bpo): Turnkey agents for document-heavy insurance operations. - [Legal](https://www.context.ai/solutions/legal): Turnkey agents for the matter lifecycle, from research to production. ## Help Center Answers at https://www.context.ai/help, one page per question (FAQPage structured data on each): ### [Getting started](https://www.context.ai/help/getting-started) - [What is Context?](https://www.context.ai/help/getting-started/what-is-context): Context is a platform for building and running agents on business workflows: Workspace, Engine, Unify, and Evals, deployed separately or together. - [Who is Context built for?](https://www.context.ai/help/getting-started/who-is-context-built-for): Teams whose workflows have repeatable steps but still depend on documents, several systems, or expert judgment. - [How do we get started?](https://www.context.ai/help/getting-started/how-do-we-get-started): Start on your own at workspace.context.ai, or request a guided demo and bring one representative workflow. - [Can I start on my own without talking to sales?](https://www.context.ai/help/getting-started/can-i-start-on-my-own-for-free): Yes. Sign up at workspace.context.ai on the free plan and upgrade from Settings when you need more. - [What does the get-started page do?](https://www.context.ai/help/getting-started/what-is-the-get-started-launcher): You pick an outcome, tune the brief, sign in, review the prepared task, and choose Send. Nothing runs from the page itself. - [How do I send my first task?](https://www.context.ai/help/getting-started/how-do-i-send-my-first-task): Describe the job in plain words, pick an agent, and send. The run keeps a full record you can inspect, and follow-ups continue the same task. - [What can Context create?](https://www.context.ai/help/getting-started/what-can-context-create): Finished decks, documents, spreadsheets, images, video, code, and research, as real files your team can edit and ship. - [How is this different from one chat subscription?](https://www.context.ai/help/getting-started/different-from-one-chat-subscription): You get model choice, shared team context, connected tools, and editable deliverables instead of another isolated chat window. - [How quickly can a small team get started?](https://www.context.ai/help/getting-started/how-quickly-can-a-small-team-start): Minutes on your own. Or bring one real use case and we help configure the workspace around it. - [How long does deployment take?](https://www.context.ai/help/getting-started/how-long-does-deployment-take): It depends on deployment requirements, connector access, and the workflow's risk; Context scopes those before committing to a schedule. ### [Using Context](https://www.context.ai/help/using-context) - [How is Context different from a chatbot or copilot?](https://www.context.ai/help/using-context/different-from-a-copilot): A chatbot returns an answer. Context runs a defined sequence of steps across your systems and keeps the record of every run. - [How is Context different from vertical AI tools?](https://www.context.ai/help/using-context/different-from-vertical-tools): A vertical tool packages one workflow. Context gives you reusable runtime, context, evaluation, and governance for many workflows. - [What is a runbook?](https://www.context.ai/help/using-context/what-is-a-runbook): The operating specification for an agent: steps, systems, inputs and outputs, approval points, and evaluation criteria. - [What are organizations and workspaces?](https://www.context.ai/help/using-context/what-is-an-organization-and-a-workspace): The organization is your tenant and billing boundary. Workspaces are the collaboration and file scopes inside it, shared or personal. - [What is an agent in Context?](https://www.context.ai/help/using-context/what-is-an-agent): A named member of your organization with its own identity, instructions, model settings, skills, and connections. - [What is a task, and what states can it be in?](https://www.context.ai/help/using-context/what-is-a-task): A task is one unit of work sent to an agent. It moves through planned, active, completed, or failed, and keeps its full record. - [Can I steer a task while it runs?](https://www.context.ai/help/using-context/can-i-steer-a-running-task): Yes. Send a follow-up message and the agent takes it into account mid-run; you can also cancel at any point. - [What are automations?](https://www.context.ai/help/using-context/what-are-automations): A task on a schedule: hourly, daily, weekdays, or weekly. Each run is a normal task with its own record, and admins can review every standing schedule. - [Does Context produce real documents, spreadsheets, and decks?](https://www.context.ai/help/using-context/does-context-produce-real-documents): Yes. Word, Excel, and PowerPoint files on your Drive, opened in a browser editor with live co-editing; the agent edits them in place on request. - [What is Drive, and how do agents use it?](https://www.context.ai/help/using-context/what-is-drive): Drive is your organization's file storage: a shared area per workspace plus a personal space. Agents read only what the requester can read. - [What is Together?](https://www.context.ai/help/using-context/what-is-together): Channels and direct messages where people and agents are members. Ask in a thread and the agent answers there, and can start background tasks from the conversation. - [Do agents remember previous work?](https://www.context.ai/help/using-context/does-context-remember-things): Yes. Agents keep memory across sessions inside your organization, and it never leaves your deployment or trains models for anyone else. - [What is an applet?](https://www.context.ai/help/using-context/what-is-an-applet): A small persistent app an agent builds for a workflow, such as a dashboard, calculator, or review tool, that opens in the workspace panel. - [What does Context Desktop do?](https://www.context.ai/help/using-context/how-do-i-use-context-desktop): Context Desktop is the Context app for Mac, Windows and Linux: your workspaces and your own deployments in tabs, with Cascade dictation into any app. It is available to download now. - [Is there a Context app for my phone?](https://www.context.ai/help/using-context/is-there-a-mobile-app): Context Mobile brings your workspace to iPhone, with notifications you can act on, a Live Activity for running tasks and home screen widgets. It is in private preview. - [How do I install the CLI and pair it with my workspace?](https://www.context.ai/help/using-context/how-do-i-install-the-cli): One install command, then context-code login prints a pairing code you approve in the workspace. Sessions land in your task list. ### [Building agents, skills, and applets](https://www.context.ai/help/building) - [How do I create an agent?](https://www.context.ai/help/building/how-do-i-create-an-agent): From a template or from scratch: give it a name, instructions, model settings, skills, workspaces, and connections. - [How do I write and share a skill?](https://www.context.ai/help/building/how-do-i-write-a-skill): A skill is a folder with a SKILL.md and supporting files. Every save is a numbered version; visibility goes personal, workspace, then organization. - [How do skills get better over time?](https://www.context.ai/help/building/how-do-skills-improve-over-time): Runs are scored against your rubrics, Context proposes improved skill versions from what worked, and a person approves before anything ships. - [How do I build an applet?](https://www.context.ai/help/building/how-do-i-build-an-applet): With the context-cli applet pipeline: scaffold, iterate on a draft, and deploy a single self-contained HTML file with shared state and optional backend functions. - [Is there an API?](https://www.context.ai/help/building/is-there-an-api): Yes. A REST Task API dispatches work to agents (files, runs, results, cancel, webhooks), a read-only Evals API exposes scores and traces, and an MCP server exposes workspace tools. - [What kinds of API keys exist?](https://www.context.ai/help/building/what-api-key-types-exist): Agent keys authenticate as an agent; task keys dispatch work with capability scopes; paired CLI keys belong to one device. Secrets are shown once and every key can be revoked. - [How do I revoke one laptop's CLI access?](https://www.context.ai/help/building/how-do-i-revoke-a-device): Each paired device holds its own named key. Revoke that key in the workspace and only that device loses access. - [What is in the marketplace?](https://www.context.ai/help/building/what-is-in-the-marketplace): Prebuilt agents across nine industries, templates, applets, skills, and a catalog of 180+ integrations, as starting points you adapt to your process. ### [Integrations and connectors](https://www.context.ai/help/integrations) - [How do agents access our systems?](https://www.context.ai/help/integrations/how-do-agents-access-our-systems): Each agent runs in an isolated environment with only the tools its runbook allows; connectors use OAuth or a scoped service credential. - [Do agents hold credentials?](https://www.context.ai/help/integrations/do-agents-hold-credentials): No. Credentials are brokered to the connector at runtime and never written into the runbook or prompt. - [Which systems can Context connect to?](https://www.context.ai/help/integrations/which-integrations-exist): A catalog of 180+ integrations across data, documents, communication, CRM, ITSM, finance, and identity, plus any remote MCP server you register. - [How does a connector authenticate?](https://www.context.ai/help/integrations/how-do-connectors-authenticate): Through OAuth on a hosted consent page or an API key you enter there. Connections are per agent or per user, and tokens never enter the sandbox. - [Can we add our own MCP server or internal tool?](https://www.context.ai/help/integrations/can-we-add-our-own-mcp-server): Yes. An organization admin registers a remote MCP server as a connector, users authorize it with OAuth, and its tools are discovered live. - [Does Context work with Word, Excel, and PowerPoint files?](https://www.context.ai/help/integrations/does-context-work-with-office-files): Yes. .docx, .xlsx, and .pptx files are created and edited in a full browser editor inside the workspace and download as standard Office files. - [What outbound network access does a self-hosted deployment need?](https://www.context.ai/help/integrations/what-outbound-access-does-a-deployment-need): The model endpoints you allow, the connector service if connectors are on, and the release proxy for online installs. Sandbox egress is a posture you set. ### [Deployment and infrastructure](https://www.context.ai/help/deployment) - [Where does Context run?](https://www.context.ai/help/deployment/where-does-context-run): Managed by Context, in your own VPC, on-premises, or in an air-gapped environment. Endpoints, connectors, updates, and support are configured for the architecture you select. - [What data leaves our boundary?](https://www.context.ai/help/deployment/what-leaves-the-boundary): Run traces stay inside your deployment. In a VPC, on-premises, or air-gapped deployment the whole platform runs inside your perimeter; model inference is the one documented cross-boundary flow. - [Is the whole platform inside our perimeter, or just inference?](https://www.context.ai/help/deployment/entire-platform-inside): The entire platform. Control plane, execution, and run history deploy in your VPC, on-premises, or air-gapped, not only model inference. - [Which clouds are supported for a VPC deployment?](https://www.context.ai/help/deployment/which-clouds-are-supported): AWS is the reference deployment, delivered to your EKS cluster through Replicated or Helm. An Azure edition is documented. Ask about other clouds. - [How is a self-hosted install delivered?](https://www.context.ai/help/deployment/how-is-a-self-hosted-install-delivered): Through the Replicated admin console (license, preflight checks, a short configuration form) or Helm from the umbrella chart. Upgrades are an operator action. - [What do we need in our account before installing?](https://www.context.ai/help/deployment/what-are-the-infrastructure-prerequisites): A Kubernetes cluster with a load balancer controller, storage and ingress classes, and an IAM role; PostgreSQL with pgvector, S3-compatible storage, and a document server. - [Can Context run fully air-gapped?](https://www.context.ai/help/deployment/can-context-run-air-gapped): Yes. Release bundles and images are side-loaded into your registry, sandbox egress is closed or allow-listed, and open-weight models are served inside the boundary. - [How do updates reach a self-hosted deployment?](https://www.context.ai/help/deployment/how-do-updates-reach-us): Connected installs see a new version in the admin console and an operator deploys it. Air-gapped installs side-load release bundles. Nothing updates without your action. - [How are agent sandboxes isolated?](https://www.context.ai/help/deployment/how-are-agent-sandboxes-isolated): Every task run gets its own sandbox under a dedicated controller, with a network posture you set and no standing credentials inside it. The sandbox is discarded after the run. - [What monitoring ships with a deployment?](https://www.context.ai/help/deployment/what-observability-is-included): An in-cluster observability stack (metrics, logs, traces, dashboards), trace analytics, and health endpoints. External export is opt-in. - [Who operates what in each deployment model?](https://www.context.ai/help/deployment/who-operates-what): Managed: Context operates, you govern policy. Your VPC: your cloud controls, operations shared per the design. On-premises and air-gapped: you operate. ### [Security, identity, and access](https://www.context.ai/help/security-and-access) - [What does the audit story look like?](https://www.context.ai/help/security-and-access/what-does-the-audit-story-look-like): Every run records the task, model and tool calls, sources, actions, approvals, and result in an append-only trail that includes every grant and denial. - [How do we get security documentation for a review?](https://www.context.ai/help/security-and-access/request-documentation): Use the request form in the Trust Center or email security@context.ai. Public documents link directly; NDA documents are shared once it is in place. - [Do you have SOC 2 or ISO 27001?](https://www.context.ai/help/security-and-access/what-certifications-do-you-hold): Yes. Request the SOC 2 and ISO 27001 reports from our Trust Center. - [Is the platform penetration-tested?](https://www.context.ai/help/security-and-access/is-the-platform-penetration-tested): Yes, by a third party as part of the vulnerability-management program committed to in the DPA. The latest report is available on request under NDA. - [How is data encrypted?](https://www.context.ai/help/security-and-access/how-is-data-encrypted): TLS in transit and industry-standard encryption at rest, as one item in the written security program in the DPA's Annex 2. - [How are tenants isolated?](https://www.context.ai/help/security-and-access/how-is-tenant-isolation-enforced): Your organization is a predicate bound into every query, not a filter applied afterwards, so another tenant's row is never a candidate result. It is identical in all four deployment models. - [How is each agent action authorized?](https://www.context.ai/help/security-and-access/how-does-per-action-authorization-work): Per action, not per session: deny, then allow, then refuse, with the specific resource bound into the decision. Read, write, and human approval are configured separately. - [How are applets kept safe?](https://www.context.ai/help/security-and-access/how-are-applets-sandboxed): An applet renders in an isolated frame with no network access and reaches the workspace only through a narrow bridge that needs a real user click. Backend functions run isolated too. - [Can agents reach the internet from a sandbox?](https://www.context.ai/help/security-and-access/what-is-the-sandbox-egress-policy): Only as your deployment allows: open, a domain allow-list, or workspace-only. The posture is stated to the agent, and connectors go through the broker either way. - [How do I report a vulnerability?](https://www.context.ai/help/security-and-access/how-do-i-report-a-vulnerability): Email security@context.ai with the affected surface, reproduction steps, and the impact you observed. Remediation and disclosure timing are coordinated in that thread. - [How quickly do you notify a data breach?](https://www.context.ai/help/security-and-access/what-is-the-breach-notification-commitment): Within 48 hours once a suspected breach is judged likely, within 72 hours once confirmed, and a written summary within ten business days of resolution, per the DPA. - [Do you support single sign-on?](https://www.context.ai/help/security-and-access/do-you-support-sso): Yes. SAML and OIDC are configured in the deployment's setup console with a test sign-in before activation; password login can stay on or be turned off. - [Which identity providers work with Context?](https://www.context.ai/help/security-and-access/which-identity-providers): Okta and Microsoft Entra ID are the named integrations; any standard SAML or OIDC provider configures the same way. - [What roles exist?](https://www.context.ai/help/security-and-access/what-roles-exist): Organization roles are owner, admin, and member (agents hold one too). Workspace roles are owner, admin, member, and viewer, with custom roles available. - [How do we invite people?](https://www.context.ai/help/security-and-access/how-do-invitations-work): Owners and admins invite by email and can pre-assign workspaces and roles. Deployments can require an invitation to join at all. - [What happens when someone leaves?](https://www.context.ai/help/security-and-access/what-happens-when-someone-leaves): Removing their membership ends their sessions, revokes their keys, and stops any schedules or triggers that ran on their authority. - [Can an admin see my personal workspace?](https://www.context.ai/help/security-and-access/can-admins-see-personal-workspaces): No. A personal workspace never gains another human member, and an administrator role confers nothing inside it. - [How do agents get their identity?](https://www.context.ai/help/security-and-access/how-do-agents-get-identity): Each agent is its own principal with an organization role. When a run starts it is bound to the agent, the requester, and the policy, and its permissions derive from the people it acts for. ### [Data, privacy, and models](https://www.context.ai/help/data-and-models) - [Is our data used to train models?](https://www.context.ai/help/data-and-models/is-our-data-used-to-train-models): No. One customer's traces, corrections, and context are never used to train models for other customers. Provider settings depend on the endpoints you choose and are verified at deployment. - [Which model providers are involved, and what do they see?](https://www.context.ai/help/data-and-models/which-model-providers): Context is model-agnostic: Claude, GPT, Gemini, Kimi, or open weights. Which endpoints are reachable depends on your deployment, and each provider's retention settings are documented with you. - [Which models does Context use?](https://www.context.ai/help/data-and-models/which-models-does-context-use): Commercial or open-weight models available in your deployment. A runbook can pin one model or route between models, and providers can be swapped without rebuilding the workflow. - [Can I switch models mid-task?](https://www.context.ai/help/data-and-models/can-i-switch-models-mid-task): Yes. Your files, connected tools, and prior work stay in place, so you can change models without starting over. - [Can we bring our own model keys or private endpoints?](https://www.context.ai/help/data-and-models/can-we-use-our-own-model-endpoints): Yes. Each deployment has an explicit provider allow-list, and a private inference gateway keeps model traffic inside your network. Your own provider account is not a subprocessor. - [How does Context learn how our company works?](https://www.context.ai/help/data-and-models/how-does-context-learn): Unify keeps your procedures, accepted examples, and corrections in a filesystem agents navigate. Proposed updates from completed work go through the review policy your team sets. - [How do we know the work is good?](https://www.context.ai/help/data-and-models/how-do-we-know-the-work-is-good): Your team writes task-specific rubrics and accepted examples; Evals scores runs against them, surfaces failures, and compares changes. Humans stay in the loop for high-consequence decisions. - [How do evals work?](https://www.context.ai/help/data-and-models/how-do-evals-work): Scorecards, judges, datasets, benchmarks, traces, and dashboards in one suite. Every run can be scored, and a read-only API exposes the results. - [What data does Context store?](https://www.context.ai/help/data-and-models/what-data-does-context-store): Five classes: institutional context, run traces, customer content, session recordings, and telemetry. Sensitive Data is out of scope under the DPA. - [Where does our data live?](https://www.context.ai/help/data-and-models/where-does-our-data-live): Inside your deployment in every mode. In a VPC, on-premises, or air-gapped deployment that means inside your own perimeter; model inference is the only cross-boundary flow. - [How long is our data retained?](https://www.context.ai/help/data-and-models/how-long-is-data-retained): Operational retention is configured per deployment. Under the DPA, data is deleted or returned within 30 days of your election when the agreement ends, backups excepted. - [Can we export our data?](https://www.context.ai/help/data-and-models/can-we-export-our-data): Yes. Drive folders download as ZIP archives and single files directly; skills export through the CLI; evals data is available through the read-only API. Return or deletion at exit is a DPA commitment. - [Who are your subprocessors?](https://www.context.ai/help/data-and-models/who-are-your-subprocessors): The current list, with purpose and region for each, is published on the subprocessors page and in the Trust Center. A provider account you supply yourself is not a subprocessor. - [How are subprocessor changes notified?](https://www.context.ai/help/data-and-models/how-are-subprocessor-changes-notified): With at least ten days' notice and a written objection right; if an objection cannot be accommodated you may terminate the affected service with a pro-rata refund. - [How are international data transfers handled?](https://www.context.ai/help/data-and-models/how-are-international-transfers-handled): Processing takes place in the United States; the Standard Contractual Clauses and the UK addendum are incorporated where GDPR, UK, or Swiss law applies. ### [Plans, billing, and procurement](https://www.context.ai/help/pricing-and-procurement) - [How is Context priced?](https://www.context.ai/help/pricing-and-procurement/how-is-context-priced): Enterprise deployments are quoted by deployment model, usage, support requirements, and implementation scope. Self-serve workspaces start free and upgrade in-product. - [How does self-serve billing work?](https://www.context.ai/help/pricing-and-procurement/how-does-self-serve-billing-work): Plans include a monthly allowance of usage credits and a number of concurrent agents. Model usage draws down credits, and you can top up or upgrade from Settings, Billing. - [What happens when credits run out?](https://www.context.ai/help/pricing-and-procurement/what-happens-when-credits-run-out): The next turn is refused until you top up or upgrade. Nothing already produced is lost, and the billing page warns before you reach the floor. - [Is the CLI billed separately?](https://www.context.ai/help/pricing-and-procurement/is-context-code-billed-separately): No. context-code has no subscription of its own; each session is a workspace task and bills as usage to your account. - [What are the enterprise payment terms?](https://www.context.ai/help/pricing-and-procurement/what-are-the-payment-terms): Fees are per order, in US dollars, due net 30, and non-refundable. Terms auto-renew unless either party gives 30 days' notice. - [Which procurement documents can we get?](https://www.context.ai/help/pricing-and-procurement/which-procurement-documents-exist): The DPA, terms, acceptable use policy, privacy policy, and subprocessor list are public. A W-9 and evidence of insurance are available on request. - [Who owns inputs and outputs?](https://www.context.ai/help/pricing-and-procurement/who-owns-inputs-and-outputs): You do. Customer data and the outputs agents produce belong to the customer; Context uses them only to provide the service. Usage data about how the service runs is Context's. - [Where are your legal documents?](https://www.context.ai/help/pricing-and-procurement/where-are-the-legal-documents): Terms, privacy policy, DPA, acceptable use policy, and the subprocessor list are published on the site and indexed on the legal page. - [What does the acceptable use policy prohibit?](https://www.context.ai/help/pricing-and-procurement/what-does-the-acceptable-use-policy-prohibit): Unlawful, deceptive, or infringing use, exceeding usage limits, and competitive scraping. For AI specifically: no training competing models on outputs, no passing outputs off as human, and human review for legally significant decisions. ### [Support and company](https://www.context.ai/help/support) - [What does implementation look like?](https://www.context.ai/help/support/what-does-implementation-look-like): Bring one workflow. We map its systems, permissions, and quality checks, choose the deployment model with you, and define the path to production. Enablement services are agreed per order. - [What support is included?](https://www.context.ai/help/support/what-support-is-included): Basic support through Context's standard channels during business hours is part of every agreement. Enterprise support arrangements are set per order. - [How do I contact the team?](https://www.context.ai/help/support/how-do-i-contact-support): hello@context.ai for general questions, sales@context.ai for commercial ones, security@context.ai for security, privacy@ and legal@ for those topics, or the contact form. - [How do I request a demo?](https://www.context.ai/help/support/how-do-i-request-a-demo): Use the guided-demo form and bring one workflow. The demo runs it end to end, shows the permissions and the trace, and maps the path to your environment. - [Do you offer training or enablement?](https://www.context.ai/help/support/do-you-offer-training): Yes, by agreement. Enablement services are scoped in the order, and on-site work is billed with reasonable expenses. - [Who makes Context?](https://www.context.ai/help/support/who-makes-context): Explore Interfaces Inc., a Delaware corporation based at 55 2nd Street, Suite 1925, San Francisco, CA 94105. - [What is open source?](https://www.context.ai/help/support/what-is-open-source): Context's software includes open-source components. Their license notices ship with the software, and a component list is available on request. - [How do I exercise my privacy rights?](https://www.context.ai/help/support/how-do-i-exercise-privacy-rights): Email privacy@context.ai. The privacy policy covers access, correction, and deletion requests, US state privacy laws, and cross-border transfers. - [How do I reach press, or get brand assets?](https://www.context.ai/help/support/press-and-brand): Email media@context.ai. Wordmarks, product screenshots, and founder photos are available on request. ## Blog - [Own your intelligence: the state of enterprise AI in Q3 2026](https://www.context.ai/blog/own-your-intelligence): The model layer is commoditizing and value is migrating to the application layer. Most firms are buying the wrong one. Where value leaks, what compounds, and the sequence to a model of the firm's own. - [Applets: generate the interface](https://www.context.ai/blog/applets-generate-the-interface): Most software ships one fixed UI and asks teams to bend to it. If an agent can write code, it can write the interface: a per-use-case applet, generated in hours, that humans and agents co-inhabit. - [The graduation ratchet](https://www.context.ai/blog/the-graduation-ratchet): Deploying AI to a workflow is not a switch. A workflow climbs a ratchet, assisted to first-pass to autonomous, each notch earned by measured quality and each one a door that does not open backward. - [Sleep-time compute](https://www.context.ai/blog/sleep-time-compute): Agents generate raw exhaust as they work. Distilling it into usable context is expensive, so we do it off the hot path, on idle compute. Tomorrow's retrieval improves; no task today is slower. - [There is no benchmark for your definition of quality](https://www.context.ai/blog/no-benchmark-for-quality): Leaderboards measure generic quality. They say nothing about whether an agent did the work the way your team would. The standard that matters is a multi-dimensional rubric your experts author and score on real tasks. - [Deploying agents in your own VPC](https://www.context.ai/blog/agents-in-your-vpc): Most enterprise AI is multi-tenant SaaS, a non-starter for sensitive work. The whole platform runs inside the customer's VPC, inference leaves only over a private endpoint, and there is no phone-home. - [A filesystem for institutional knowledge](https://www.context.ai/blog/a-filesystem-for-context): The default way to give an agent a company's knowledge is a vector database. We built a filesystem the agent traverses like a code repo, with a .context directory at every level. ## More - [Marketplace: agents and integrations](https://www.context.ai/marketplace) - [Qualcomm case study](https://www.context.ai/customers/qualcomm) - [Get started: first deliverables](https://www.context.ai/get-started) - [Contact](https://www.context.ai/contact) - [Sitemap](https://www.context.ai/sitemap.xml)