HQ by Indigo

A one-sentence summary

HQ by Indigo is an AI operating system designed for enterprises and teams; it stores knowledge, skills, projects, rules, and work records in searchable documents, and then provides this context to coding assistants, chat tools, and persistent intelligent agents.

Tool Introduction

HQ was developed by Indigo AI; it is not intended to be a new large language model or a standard chatbot, but rather a layer that provides a shared context and capabilities between the models and the operations of an enterprise. Teams can continue to use the AI tools they are familiar with, while simultaneously reducing the need to re-explain business practices, processes, and past decisions in each conversation.

The entire product suite consists of HQ Core, HQ Cloud, Secrets and Integrations, HQ Deploy, HQ Agents, and HQ MCP. The open-source core is responsible for file-based memory and the execution framework, while the hosting services add capabilities for synchronization, permissions, keys, connectors, resident agents, and deployment.

Product composition

Main functions

Document-based corporate memory

HQ Core uses ordinary files to store corporate knowledge, projects, personnel information, decisions, skills, and policies. Members can view, edit, compare, and transfer this information directly, without having to keep their long-term memories confined to a single conversation or a particular modeling service.

Session context loading

The execution framework identifies the current company and project, and then feeds the relevant memories, rules, and capabilities into the AI session. The accuracy of the context depends on the quality of file maintenance; it is necessary to have someone responsible for updating expiration policies and conflict records.

Sharing skills and workflows

The team can turn repetitive tasks into skills or slash commands, including research, content creation, data analysis, code review, and release checks. A process that has been verified by a team member can be made available to other members for use, and it can be reviewed and rolled back through version tracking of files.

Workers can reuse experts.

A Worker is a file-based expert that can be called as needed within active AI sessions; it may include instructions, skills, scripts, and task queues. It is suitable for carrying out specialized tasks within a particular process, but it does not have its own Slack or email account, nor does it inherently serve as the person in charge of ongoing tasks.

HQ Agents resident agents

An Agent has a fixed name, role, membership in the company, memory, authorized tools, and its own operating environment. It can receive tasks via Slack or email, or it can initiate tasks on a scheduled basis or in response to company events.

Division of roles between Agent and Worker

Agents are more suitable when long-term responsibility, a unified identity, and cross-channel communication are required, while Workers are better suited for invoking a professional executor within the current session. Both can perform external operations; therefore, high-risk processes should be configured with minimal permissions and require manual confirmation.

Team synchronization and conflict resolution

HQ Cloud can synchronize approved knowledge, projects, skills, Workers, policies, connections, and other shared content. Changes that cannot be merged safely are sent to members for review, rather than silently overwriting one of the versions.

Enterprise Console

Administrators can browse the shared Vault in the console, manage administrators, Agents, groups, Secrets, connections, and access permissions, as well as monitor team usage. The console serves as a management and visibility layer; it does not replace the file-based HQ workspace on members’ computers.

Key and credential management

Secrets stores API keys and login credentials in an encrypted vault, preventing members from pasting the actual values in prompts, documents, chat records, or code repositories. Administrators can grant access to specific individuals, groups, or Agents for certain credentials without releasing the original values.

Runtime key injection

The project can specify which keys are required through an environment variable configuration, and HQ will then insert the approved values when the commands are executed. The actual keys are not printed out, but malicious scripts, excessive permissions, or error logs can still lead to the exposure of credentials; therefore, code review and key rotation must not be neglected.

System integration and custom connections

A team can connect to multiple business systems or custom MCP services at once, and then provide these connections to sessions and agents based on their permissions. The level of functionality offered by different integrations varies; some only support queries, some allow for performing actions, while others require separate development.

HQ MCP chat integration

HQ MCP provides enterprises with their knowledge, skills, tools, and connections, making them available to chat interfaces such as Claude and ChatGPT. Regular members can continue to use the familiar conversation methods, but they can only access the enterprise resources to which their accounts have been granted access.

Semantic and keyword search

HQ allows for searching within knowledge bases, projects, policies, code, and work records; it supports keyword-based, semantic, and mixed search methods. The quality of the search results depends on the file structure, the status of the indexing, and the metadata, and it is still necessary to consult the original records for important decisions.

Project planning and automatic execution

The open-source workspace provides commands and workers for planning, assigning, executing, and checking tasks, allowing complex jobs to be broken down into multiple steps. Automatic loops are suitable for tasks with clear boundaries and verifiable outcomes; permissions should not be granted indiscriminately in situations involving payments, deletions, publications, or customer communications.

Session handover and learning

At the end of a task, it is possible to save threads, submissions, modified files, and subsequent steps, so that the work can be continued from where it left off in the next session. The team can also record users’ corrections and task-related experiences in the relevant skill or knowledge documents, but it is necessary to first assess whether such information is applicable to a broader range of situations.

Deployment of AI products

HQ Deploy allows AI-generated reports, dashboards, portals, or lightweight applications to be published from local workspaces as pages that can be opened. The access to these published items can be set to be public, password-protected, or available only to authenticated enterprise members.

Meeting capture and enterprise signals

The team solution includes meeting and corporate intelligence capabilities; it allows for the recording of meetings, the generation of transcripts and summaries, as well as the identification of decisions, action items, and key facts. Meeting recording is charged on an hourly basis, and organizations also need to handle matters related to participant consent, audio recording alerts, retention periods, and access to sensitive discussions.

AI usage visibility

Managers can observe which skills are being utilized, which tasks are reused, and how people and Agents work together. Statistics are useful for identifying the frequency of certain practices and repetitive processes, but they should not be used simply to assign scores to individual performance.

Isolation of multiple companies

Each company, acting as an independent trust boundary, possesses its own knowledge, members, policies, connections, and credentials. Agencies or multi-brand teams still need to place clients in the appropriate work areas, and regularly check for imports, copies, and authorizations across companies.

Typical usage process

1. Install the desktop version or use the command line to create an HQ workspace.
2. Create or join a company, and verify that the current workspace corresponds to the correct organization.
3. Run the setup wizard to add information regarding the company’s background, policies, personnel, and project structure.
4. Connect to coding assistants such as Claude Code, Codex, or Cursor that offer support.
5. Enable Cloud synchronization and assign only the minimum necessary permissions to members and groups.
6. Save the API key in Secrets; it is not allowed to paste the actual value in chats or ordinary files.
7. Create the first shared skill or Worker, and verify it using a representative task.
8. Configure HQ MCP when chat access is required, and create an Agent when a permanent owner is needed.
9. Save decisions, documents, task status, and next steps through handover commands.
10. Regularly review expired knowledge, connection permissions, keys, deployment pages, and agent activities.

Suggestions for implementing the team approach

1. First, select a business process whose outcomes can be quantified, whose risks are low, and that is carried out frequently.
2. Designate responsible persons for corporate knowledge, skills, connections, and Agents respectively.
3. Organize the current policies and standard answers, and remove any contradictory or outdated materials.
4. Create groups based on positions, and grant read-only or single connection permissions first.
5. Run the skills in the testing workspace, and record the successful conditions, failed samples, and manual inspection points.
6. The Agent is allowed to publish, send messages, or modify external systems only after the rollback method has been confirmed.
7. Gradually make chat, Agent, and access to deployed applications available to non-technical users.
8. Review the call costs, meeting hours, as well as the usage of additional agents and Outposts on a monthly basis.
9. Randomly check the AI-generated conclusions against the actual business records to prevent errors from spreading rapidly within the team.
10. Retain the exportable files and migration plans, and conduct regular drills for recovery and exit procedures.

Prices and packages

The price information was verified on August 23, 2026; the actual amounts, taxes, exchange rates, and discounts may vary, and the final figures will be those displayed on the settlement page.

HQ utilizes a company package that does not charge for seats; the number of employees is generally not subject to any seating fees. Subscriptions for external AI models, costs associated with invoking those models, and fees related to third-party services are not included in the HQ price. The actual bill will also be affected by regional factors and taxes.

Startup Free Version

The free version supports up to 50 users, charges no fee per seat, and provides Vault, synchronization, Secrets, integration capabilities as well as a limited number of deployment slots. The official pricing page lists Agents, meetings, Ontology, and MCP options, but the actual scope of free usage may be subject to eligibility requirements and usage limits.

Team Team Edition

The Team version comes at a fixed price of $500 per month per company; it can be canceled on a monthly basis and includes 3 AI Agents. The White Glove setup service entails an assessment of the AI readiness, two training sessions for the team, one day of on-site activation, as well as dedicated support channels.

Enterprise edition

The enterprise version starts at $10,000 per month and offers SSO, dedicated infrastructure, and corporate support. The specific number of users, data storage location, SLAs, auditing requirements, retention periods, procurement timelines, and implementation services are subject to confirmation via contract.

Pay-as-you-go pricing

The authorities provide metrics for usage and budget limits, but the contents of the packages may still change. Before making a purchase, it is necessary to determine whether a credit card is required, the starting date, the rules regarding refunds, taxes, the number of additional agents, and the budget allocated for meeting hours.

Supported platforms and access points

Which users are it suitable for

• Teams that wish for different AI tools to share the same set of corporate knowledge and policies.
• Development organizations that are already using Claude Code, Codex, or Cursor for actual work.
• An operations team that can turn success messages and procedures into shared skills is needed.
• The goal is to enable non-technical members to use companies with approved corporate contexts through chat.
• Companies that need to deploy permanent AI teammates via Slack or email.
• Platform teams that need to manage interface keys, MCP connections, and Agent permissions in a centralized manner.
• Teams that frequently create internal dashboards, reports, portals, and lightweight applications.
• Individual developers who wish to save AI working memory in the form of transferable ordinary files.

It is not very suitable for which users

• Individual users who simply want to have a universal chat robot ready for use out of the box.
• Organizations that have no staff responsible for maintaining corporate knowledge, policies, and permissions.
• Teams that require the entire hosting platform to be open source and available for free self-hosting.
• Businesses that do not allow AI tools to access any corporate documents or external connections are not acceptable.
• Teams that lack the budget to purchase external model subscriptions but wish to run a large number of agents.
• Organizations that hope to enable agents to make payments, delete data, or contact customers independently, without the need for review.
• Companies that are reluctant to use Git, file systems, or workspaces maintained by small teams of operators.
• Regulated entities that require certification from mature industries but have not yet completed contract and security due diligence.

Product advantages

• Save the company’s history in ordinary files, making it easy to view, edit, version, and transfer.
• It is not tied to a single model; it can be connected to various encoding assistants and chat tools.
• Integrate knowledge, skills, Workers, Agents, and deployments into a unified working system.
• The free Startup plan supports up to 50 people, and there is no charge based on the number of employees.
• Secrets reduces the risk of plaintext exposure through separate encryption and runtime injection.
• It supports granting permissions for files, connections, and credentials on a per-user, per-group, and per-Agent basis.
• Sharing skills allows individual-tested working methods to be turned into team standards.
• HQ Core is licensed under the MIT license, allowing developers to examine and modify its core structure.
• Synchronization conflicts are checked by the members; files are not silently overwritten by default.

Usage restrictions and precautions

• HQ is not a base model; it must be used separately and comes with the cost of compatible AI services.
• Document-based memory does not guarantee accuracy automatically, and incorrect information can also be used by the entire team.
• Local priority does not mean complete offline operation; cloud synchronization, MCP, and Agents will handle the approved data.
• Although the startup version is free, there are clear limits on Secrets, storage, integration, and deployment.
• Additional Agents, Outposts, and meeting recordings outside of Team will continue to incur costs.
• The enterprise version has a higher starting price, and SSO as well as dedicated infrastructure need to be purchased separately.
• The depth of queries and operations supported by different integrations varies, and each one needs to be verified individually.
• Runtime injection of keys cannot prevent malicious code or faulty tools from exporting credentials.
• Resident Agents may initiate tasks on their own, and it is necessary to restrict the tools, identities, and triggering conditions.
• Both public and password-protected deployments can still lead to the leakage of sensitive reports; therefore, the appropriate access patterns must be used.
• There may be differences in cross-platform installation and functional consistency, and separate testing is required for server environments.
• AI cannot completely eliminate hallucinations; key conclusions, code, and external operations require manual verification.

Security, privacy, and data processing

HQ Core stores working memory in local files that can be accessed by the team; with HQ Cloud, the data that is authorized for use in cloud-based functions is synchronized across all tenants within the company. Cloud data is encrypted both during transmission and when stored statically, and sensitive credentials are encrypted at the field level as well.

Cloud infrastructure

According to the official specifications, meetings and documents are stored in AWS S3, while application data is stored in MongoDB Atlas; encryption is applied both during transmission and at rest. Organizations should further verify the selection of regions, backup procedures, sub-processors, procedures for incident notification, recovery objectives, and evidence of data deletion after contract termination.

Meetings and Google data

The privacy policy specifies that meeting audio, transcripts, metadata, as well as content from Gmail, Drive, Docs, Sheets, Calendar, and Contacts may be processed. Once Google’s authorization is revoked or a specific request for deletion is made, the relevant data is usually deleted within 30 days; however, financial, legal, audit-related data, and anonymous data may be retained as an exception.

Permissions and high-risk operations

HQ can grant permissions for files, Secrets, integrations, MCP, deployment, and Agent capabilities to individuals, groups, and Agents respectively. Skills related to publishing content, making financial transactions, modifying the system, or contacting customers require more stringent testing, approval, and manual verification.

Open-source status and developer capabilities

HQ Core is available on the official GitHub repository under the MIT license; developers can create workspaces using specific commands, or they can install the official command-line tools. The public repository contains configurations for tools such as Claude and Codex, as well as the directory structures for companies, individuals, projects, and the core components.

The scope of open sourcing applies mainly to the file-based memory and execution framework of HQ Core; it does not mean that HQ Cloud, Secrets, Agents, MCP, Deploy, and the enterprise console are all open sourced. Companies that need to deploy these services privately or make modifications to the hosted services should consult the vendor regarding code delivery, licenses, infrastructure, and maintenance responsibilities.

Basic information

Frequently Asked Questions

What is HQ by Indigo?

It is an operating system that consolidates a company’s knowledge, skills, projects, policies, keys, connections, and AI agents. It provides ongoing context to existing models, rather than replacing them.

Is HQ free?

Startup is available free of charge for teams of up to 50 members, and HQ Core is also open-source under the MIT license. The Team version, Enterprise version, additional Agents, Outpost, and meeting recording require payment.

What are the limitations of the free version?

The Startup plan includes limits such as 10 Secrets, 10 GB of storage, 5 integrations, and 500 lifetime deployments. Community support and certain advanced features also differ from those available in the paid plans.

Is it necessary to purchase a seat for each employee?

The official pricing does not include a charge based on the number of employees; Team charges a fixed fee of $500 per company per month. The ongoing costs stem mainly from packages, agents, Outpost, meetings, and external AI services.

Can HQ replace ChatGPT or Claude?

It’s not possible. HQ is responsible for the company’s memory, processes, permissions, and connections; actual generation and reasoning are still carried out by AI tools such as Claude, ChatGPT, Codex, and Cursor.

Can non-technical users use it?

Yes. Regular members can utilize the approved capabilities through Claude, ChatGPT, Slack, email, or the deployed pages, but usually a team of operators is still needed to manage the local HQ, skills, integrations, and permissions.

Does HQ need Git or GitHub?

The full local workspace relies on Git as a technical tool for tracking history, conducting reviews, and facilitating collaboration; however, it is not required that non-technical members manage branches directly. HQ also does not automatically make corporate files available on GitHub.

Does HQ support Windows and Linux?

The official version offers installation options for macOS and Windows, while the underlying file-based workspace can also be used in Git Bash environments on Linux, macOS, and Windows. The installation methods and functional capabilities may vary depending on the platform.

Will HQ use corporate meetings to train the model?

The privacy policy states that meeting contents, transcripts, and extracted signals are used solely to provide inference results to the current user and organization, and not for training or fine-tuning models. Other corporate data must be examined separately, in accordance with the terms of contracts and the policies related to the model services in question.

Can HQ eliminate AI hallucinations?

It’s not possible. Controlled corporate documents, policies, and tools can help reduce the lack of context, but incorrect knowledge, flawed reasoning based on models, and external data can still lead to errors; therefore, it is essential to keep records and carry out manual reviews for critical tasks.

Is HQ completely open source?

No. HQ Core is available under an open-source license based on the MIT license, whereas Cloud, Secrets, Agents, MCP, Deploy, and the enterprise console belong to the commercial product suite.

Where is the best place to start with a pilot project?

One can start by choosing a low-risk repetitive process, and use Startup or the local Core to develop knowledge, skills, and inspection standards. Once the results, permissions, and costs have been confirmed, the approach can be expanded to include MCP, Agents, meetings, and operations with external systems.

Summary

HQ by Indigo is suitable for teams that are already using various AI tools in their work and wish to establish a shared corporate memory, reuse processes, and manage the permissions of intelligent agents. Its features include a document-based context, model independence, as well as a complete set of functionalities that go from knowledge to connections, including agents and deployments.

When making a selection, it is necessary to evaluate both open-source cores and hosted suites separately, paying close attention to responsibilities for knowledge maintenance, costs associated with external models, free usage quotas, Agent permissions, data related to meetings and Google, cross-border processing, as well as enterprise-level contracts. First, use a low-risk process to assess the reuse rate, accuracy, amount of manual work required, and overall cost, before deciding whether to upgrade to a team or enterprise solution.