Data retention policy
Hone retains organization-scoped product data, conversation records, engine
activity, event history, and generated artifacts while the customer account
remains active or as required by the applicable customer agreement. Hone does
not use Slack data to train LLMs. Data is removed following an authorized
deletion request or customer offboarding procedure, subject to
backup-expiration and legal-retention requirements.
Data archiving and removal policy
Following an authorized customer deletion or offboarding request, Hone deletes
organization-scoped records from PostgreSQL, ClickHouse, S3, and Redis,
including S3 object versions and delete markers. Deleted data may remain
temporarily in backups until they expire: up to 7 days for RDS and ClickHouse
backups and 1 day for Redis snapshots. Observability records expire according
to the applicable monitoring-provider retention period.
Data storage policy
Hone stores customer data in AWS us-east-1. Organization and conversation
records are stored in Amazon RDS for PostgreSQL; runtime event history is stored
in self-hosted ClickHouse on AWS; workspace files and artifacts are stored in
Amazon S3; and live event streams are stored in Amazon ElastiCache for Redis.
Data is logically scoped by Hone organization. Connector credentials and other
secrets are encrypted and stored in encrypted integration records or AWS
Secrets Manager.
Data center location(s)
United States
Data hosting details
Cloud hosted on AWS
App/service has sub-processors
yes
Guidelines for sub-processors
App/service uses large language models (LLM)
yes
LLM model(s) used
Hone uses Anthropic Claude models (including Sonnet, Opus, and Haiku) and OpenAI GPT models (including GPT-5.6 Sol, Terra, and Luna, and GPT-5 Mini). The specific model varies by engine and workflow and may change as supported model versions are updated.
LLM retention settings
Hone uses Anthropic and OpenAI APIs with Zero Data Retention enabled. For eligible API requests, the standard provider retention period for prompts and responses, including Slack content, is 0 days after processing. Exceptions may apply to specific models
LLM data tenancy policy
Hone logically isolates customer data by Hone organization. Inference requests include only the information required for the requested task. Customers do not share access to one another’s prompts, responses, or application data.
LLM data residency policy
Hone application data is hosted in AWS us-east-1. Inference data may be transmitted to Anthropic, OpenAI, or Cerebras for processing. Provider processing locations are governed by Hone’s agreements and provider configurations.