Concerns when using the system
Will there be a proliferation of agents, making it difficult to know which one to use?
Agentiqs allows you to define and share standard agents on a workspace basis.
- Standard agents can be distributed via Git or shared folders, and version control is also possible.
- Because you can clearly define the agents to be used by the organization, you can prevent a proliferation of agents and reliance on individual users.
It is also possible to operate individual agents and organizational standard agents separately.
Where is the vector database used by RAG located?
The vector database used by RAG is built on each user's local PC.
It is not a SaaS-type configuration that is located on a specific server.
Knowledge (design documents, specifications, code, etc.) is registered in folder units and vectorized and indexed on each PC.
How is team sharing achieved without a server?
Knowledge itself and agent settings are shared via Git or shared folders.
Each member incorporates this information and reconstructs the vector database on their own PC.
Are there any new processes that arise from the introduction of AI agents?
Yes, there are.
For example, the following organizational and design tasks may arise:
- Organization of review points and required check items
- Workflow design
- Review of knowledge structure
- Definition of standard agents
These are processes that formalize parts that were previously tacit knowledge and dependent on individuals.
While design and organizational work may occur temporarily, it can then be operated as a reproducible system,
leading to continuous efficiency and quality stability.
Agentiqs is designed not simply to automate tasks,
but as a foundation for structuring business processes.
Is it possible to handle confidential files without security risks?
Yes, with conditions.
First, Agentiqs is a locally running application.
- Confidential files such as design documents and code remain on your PC or within your company environment during processing.
- The vector database is also built locally and is never uploaded to our servers.
However, communication with the Generative AI (LLM) is limited to the LLM you have contracted.
By using an enterprise contract such as Azure OpenAI, it is possible to operate in a configuration where input data is not used for training.