History
Category: Agent configuration Version: 1.0 Last updated: March 26, 2026 Author: Any2Info
Description
The History & File Retention settings control how conversations and associated file references are stored and cleaned up for an agent.
Within Any2Info, it is important to distinguish between two storage layers:
Any2Info conversation storage (internal persistence)
AI provider storage (temporary or persistent, depending on configuration and mode)
A conversation represents the stored input and output of an agent execution. This applies to both chat-enabled and non-chat agents.
Chat vs Non-Chat Behavior
Chat enabled
AI provider calls are stateful. Previous messages are included in subsequent requests. Files remain available in provider storage for the duration of the conversation. Conversations are always created and stored, regardless of the history setting.
Chat disabled
AI provider calls are stateless. Messages are not chained in requests. Files are cleaned up immediately after each request. Conversations are only stored when history is enabled.
When chat is enabled and history is disabled, conversations are still stored internally but are not accessible to the user through normal means.
Conversation Storage
Chat Enabled
A conversation instance is always created
Messages are chained both:
In AI provider calls
In stored conversation data
If history is disabled:
Conversation is still stored
Conversation is not accessible to the user
Chat Disabled
A conversation is created only when history is enabled
Messages are stored and chained in the conversation
AI provider calls remain stateless (no chaining)
File Handling & Storage
Within Any2Info Conversations
Images
Full base64 content
Other files (PDF, Word, Excel, TXT, etc.)
File name and extension only
File content (except images) is not stored in Any2Info.
AI Provider Handling
Images
Sent as base64
Never stored in provider storage
Text / Content Files
If supported → uploaded to provider storage
If not supported → converted to text and sent inline
Example (OpenAI):
Only PDF files are stored in provider storage
Other file types are converted to text
File Lifecycle in AI Provider
Non-Chat Enabled Agents
Files are uploaded to provider storage during request execution
After receiving a response:
Files are immediately removed from provider storage
No persistence beyond a single request
Any2Info still stores the file reference (name + extension)
Chat Enabled Agents
Files are uploaded when provided as input
Files remain in provider storage for the entire conversation lifecycle
Files are reused across subsequent requests within the same conversation
Configuration Options
Max duration (days)
No
Defines how long a conversation may remain unused before being eligible for cleanup. The timer resets whenever the conversation is used again.
Max count
No
Defines the maximum number of stored conversations per agent. Oldest conversations are removed when the limit is exceeded.
Conversation Lifecycle
Max Count
Limits the number of stored conversations per agent
Example:
Max count = 10
When the 11th conversation is created → conversation 1 is deleted
Max Duration
Defines how long a conversation may remain unused
Based on last usage timestamp
The duration is reset whenever the conversation is used again
Example:
Max duration = 2 days
Last used: January 1, 13:18
User interacts again on January 2 → timer resets
New expiry: January 4, based on last interaction
Conversation is removed during the first cleanup cycle after expiry
The duration only applies to continuous inactive (stale) periods
Cleanup Cycles
Executed hourly
Both max duration and max count are evaluated every cycle
Configuration changes take effect immediately on next cycle
Deletion & Cleanup Behavior
When a conversation is cleaned:
Any2Info
Conversation is marked as deleted
Data remains in the database (soft delete)
AI Provider
All associated files are permanently removed from provider storage
User-triggered Cleanup
A conversation can also be explicitly ended by the user. In this case, cleanup is executed identically to the automatic cleanup process.
Ways a user can trigger this:
Using the clear messages action in an agent
Leaving a stateless (non-chat) agent conversation
Usage
Use History & File Retention to manage:
Storage lifecycle of conversations
Cleanup of AI provider stored files
Data retention policies
Typical use cases:
Limiting storage usage
Managing temporary or sensitive data
Ensuring cleanup of uploaded files
Controlling conversation lifecycle
Important Notes
File retention is fully dependent on the conversation lifecycle
There is no independent file retention mechanism
Max count indirectly affects file retention
Images are stored differently than other file types (base64 vs metadata)
AI provider storage behavior depends on:
Provider capabilities
Chat vs non-chat mode
Disabling history only affects user accessibility, not whether conversations exist (in chat mode)
Tips & Best Practices
Use Max duration for time-based cleanup policies
Use Max count to control storage volume
Prefer non-chat mode for immediate file cleanup scenarios
Use chat mode when file reuse across interactions is required
Be aware of hourly cleanup cycles when designing workflows
Changelog
1.0
March 26, 2026
Initial documentation version added.
Last updated
Was this helpful?