Deploy a Botsify AI Agent with Versioned Configurations

Learn how to configure agent instructions, save a named version, set it as active, and deploy your Botsify AI agent to production.

Botsify gives you a full version-control workflow for your AI agents. You can iterate on prompts, memory settings, and conversation flows in isolation, then promote a specific version to production with a single API call. This guide covers each step from saving a configuration through to live deployment.

1. Save an agent version

   Use POST /api/v1/update-agent to write your agent’s configuration. Every call to this endpoint creates or updates a named version associated with your bot.

   curl --request POST \
     --url https://app.botsify.com/api/v1/update-agent \
     --header 'Content-Type: application/json' \
     --header 'Authorization: Bearer <YOUR_API_KEY>' \
     --data '{
       "bot_id": "bot_abc123",
       "name": "v2-improved-tone",
       "agent": "You are a friendly support assistant. Always greet users by name.",
       "chat_flow": { "welcome_message": "Hi there! How can I help you today?" },
       "memory": "Remember the user'\''s name and previous issues across sessions.",
       "soul": "Empathetic, concise, never dismissive.",
       "identity": "Acme Support Bot — powered by Botsify.",
       "user_info": "Collect: name, email, order number.",
       "version_id": "v2-improved-tone",
       "edit": false
     }'
   

   Request body fields

   Field	Type	Required	Description
   bot_id	string	✅	The ID of the bot to update
   name	string	✅	Human-readable label for this version
   agent	string	✅	The system prompt or instructions given to the AI
   chat_flow	object	✅	Structured conversation flow definition
   memory	string	❌	Instructions for what the agent should remember across turns
   soul	string	❌	Personality and tone guidelines for the agent
   identity	string	❌	How the agent should describe itself to users
   user_info	string	❌	Fields the agent should collect from the user
   version_id	string	❌	Identifier for this version; auto-generated if omitted
   edit	boolean	❌	Set to true to overwrite an existing version rather than creating a new one

2. Review your saved versions

   Before promoting a version, fetch the full list of saved versions for your bot to confirm the version ID you want to deploy.

   curl --request GET \
     --url 'https://app.botsify.com/api/v1/bot/agent-versions?bot_id=bot_abc123' \
     --header 'Authorization: Bearer <YOUR_API_KEY>'
   

   The response returns an array of version objects — each containing its version_id, name, creation timestamp, and whether it is currently active. Note the version_id of the version you want to promote.

3. Set the active version

   Mark a specific version as the active one. This determines which configuration is used when you trigger a deployment.

   curl --request POST \
     --url https://app.botsify.com/api/v1/set-active-version \
     --header 'Content-Type: application/json' \
     --header 'Authorization: Bearer <YOUR_API_KEY>' \
     --data '{
       "version_id": "v2-improved-tone",
       "bot_id": "bot_abc123"
     }'
   

   Request body fields

   Field	Type	Required	Description
   version_id	string	✅	The ID of the version to activate
   bot_id	string	❌	Scopes the operation to a specific bot

4. Deploy the agent

   Push the active version — or a specific version — to production with POST /api/v1/deploy-ai-agent.

   curl --request POST \
     --url https://app.botsify.com/api/v1/deploy-ai-agent \
     --header 'Content-Type: application/json' \
     --header 'Authorization: Bearer <YOUR_API_KEY>' \
     --data '{
       "version_id": "v2-improved-tone",
       "new_version_name": "Production — Jan 2025"
     }'
   

   Request body fields

   Field	Type	Required	Description
   version_id	string	❌	The version to deploy; defaults to the currently active version if omitted
   new_version_name	string	❌	An optional display name for this deployment snapshot

   You do not need to pass version_id if you have already set the active version in the previous step. Omitting it will deploy whichever version is currently marked active for the bot.

Deployment is immediate and affects all live users. As soon as deploy-ai-agent returns successfully, every new conversation on your bot will use the deployed configuration. Test thoroughly in a staging bot before deploying changes that affect your agent’s soul, identity, or core agent instructions.

Rollback strategy

If a deployment causes issues, call POST /api/v1/set-active-version with the version_id of your last stable release, then redeploy with POST /api/v1/deploy-ai-agent. Because all versions are persisted, rolling back is as fast as rolling forward.

Next steps

• Monitor performance — track message counts and engagement after a deployment using Analytics.
• Send proactive messages — reach out to users with announcements or follow-ups using Send Messages.