OpenAI AI Assistants & Agents Manager
Description
Highlights
About the Extension
OpenAI AI Assistants & Agents Manager is the central backend control center for creating, configuring and testing OpenAI-powered agents in Shopware 6. It provides a shared foundation for our AI plugin suite, so agents, instructions, tools and knowledge can be configured once and reused across multiple extensions such as "NextGen OpenAI Chatbot" and "AI Product Advisor".
Under the hood, the manager uses the OpenAI Responses API (the “Assistants” naming is legacy).
Central agent management for Shopware
Create and manage agents directly in the Shopware backend. Configure models, instructions, tools and knowledge sources in one place, test their behavior before deployment and reuse the same agent setup across compatible 5 Elements AI extensions.
Smart Search with Embeddings and a local Hybrid Search Index
The manager includes a semantic Smart Search system that combines vector embeddings with a local Shopware search index. Instead of relying only on exact keywords, search queries can match relevant content based on meaning and context. This allows agents to find the right information even when customers use different wording.
The Hybrid Search Index can include products, Knowledge Base entries and selected CMS pages. Semantic embeddings and local indexing work together to provide precise retrieval without requiring agents to know the exact product name, tag or phrasing in advance.
Transparent data handling: Search indexes and generated vectors remain inside Shopware. Only the text being indexed and search queries are sent to the configured Embedding API for semantic processing.
Automatic caching and index status handling keep search results fast and up to date. Indexing is designed for reliable operation in hosted environments, can continue safely without an available background worker and supports recovery of interrupted indexing jobs.
Self-growing Knowledge Base
The built-in Knowledge Base turns unanswered questions into reusable knowledge directly inside Shopware.
- A user asks a question and the agent searches the available knowledge.
- If no suitable answer is found, a new unresolved Knowledge Base entry can be created automatically.
- You add an Answer and optionally define aliases or exact search terms.
- Once resolved, agents can reuse the answer in future conversations.
Unresolved entries can still appear in search results as unresolved, helping agents recognize that a topic already exists. searchTerms and tagAliases provide additional control for exact phrases and related terminology.
Knowledge entries can also contain multiple source URLs. When an entry is used, the system can automatically retrieve and sanitize the linked content and include it as additional context without requiring another agent tool call.
Playground, Content Checker and transparency
Test agents before using them in customer-facing workflows. The Playground provides an interactive chat environment for validating prompts, instructions and tool calls, while the central log area helps trace AI interactions and troubleshoot unexpected behavior.
The Content Checker lets you test available tools directly in the backend. Its interface is generated from the same tool definitions the agents use, so parameters match the actual implementation and displayed tool responses are the real payloads agents receive.
- Live testing of available agent tools
- Input and parameter validation
- Inspection of actual tool responses
- Central logging and trace information
- Usage tracking per agent
The Agent Manager overview also includes a Store API Health Check for identifying blocked or unavailable storefront endpoints. An interactive Token Counter helps estimate message lengths and expected API costs while preparing content and testing workflows.
Important: The Playground runs in the context of the selected Sales Channel, including language, currency and applicable rules. With Backend mode enabled, responses include additional information intended for testing and debugging. Disable Backend mode to test behavior closer to the corresponding storefront experience.
Note: Product results in the backend Playground use a condensed representation optimized for agent testing. Storefront plugins apply their own presentation layer for customer-facing product listings.
Quality assurance and regression testing
If chat logging is enabled, completed agent conversations can be reviewed directly in the backend, including per-message token usage and aggregated totals.
Scorecards can be generated from chat threads to combine conversation context, tool metadata and performance information such as token usage, caching ratio and latency. This makes it easier to review agent behavior consistently and compare results after changes to prompts, tools or configurations.
User prompts can also be exported as a plain text list and replayed with the auto-chat utility, making repeatable regression testing possible across agent updates.
Requirements / important notes
- An OpenAI API key is required (with API credits). The key must be configured for all sales channels you use.
- This is a base plugin. Without one of our AI plugins, it does not provide standalone storefront functionality.
- If you use the older plugin “Custom GPT Sales and Support Assistant – Chatbot, Shopping Helper”, uninstall it first to avoid conflicts.
- Shopware 6.7 compatible; updated to the OpenAI Responses API.
Documentation:
https://docs.5-elements.cloud/oai-assistants-manager/
Details
- Available: English, German
- Latest update: 14 August 2026
- Publication date: 25 June 2025
- Version: 2.0.9
- Category: Administration
Resources
Reviews (0)
About the Extension Partner
5 Elements Websolutions
Partner Status
-
Shopware
Extension Partner
Details
- Ø-Rating: 1
- Partner since: 2019
- Extensions: 10
Support
- Based in: Germany
- Speaks: German, English
- Response time: Very quickly