AI Assistant Platform

01/01/2024

AI Assistant Platform
Our AI Platform implements multiple intelligent virtual assistants capable of providing information regarding customer support (pre- or post-sales phase) or internal support (HR, sales, production, technical assistance, etc.). The platform has a focus on manufacturing and industrial companies, but its versatility also makes it adaptable to companies in the tertiary sector (commercial, services, banking, insurance,...) and in the advanced tertiary sector (IT, consultancy...).

The main areas of application of the platform already implemented are the following:
  • provide answers on documentation/technical information (e.g. answering questions on product technical data sheets, price lists, assembly sheets, non-conformities, ...)
  • be used as a training platform to consult courses, tutorials (including videos), manuals of various kinds
  • be of support in identifying procedures, operating instructions, forms, etc
  • provide documents based on user requests (e.g. first aid manuals, technical data sheets, ...)
  • integration with external / B2B services (e.g. integration with Microsoft Dynamics to obtain customer order status)
 

At a technical level, the proposed solution is composed of the following elements:

  1. AI Assistant Platform: it is a web-based platform created with latest generation techniques and technologies in which basic components have been implemented such as:
  • management of related user roles (basic, admin, ..) and permissions
  • connectors to cloud AI models (OpenAI, Gemini, …) and on-premises (gpt4all, …)
  • possibility of integration with external services
  • log management, chat and user reporting
  • multilingual user interface
  • platform usage statistics
  • text to speech, speech to text
  1. AI engine: these are the Artificial Intelligence model(s) that are connected to the AI Assistant Platform via dedicated adapters, making them usable and exposed to queries from the outside (end users, back-office, third-party application services, ERPs, etc.). The choice of these models is proposed by us based on the specific use case submitted to us and subsequently validated together with the customer on the basis of parameters such as goodness of the results, effectiveness of the model, licensing/engine prices.

The sales model of the solution includes the following:
  1. license to use the AI Assistant Platform with an annual fee which includes corrective, adaptive and evolutionary maintenance of the platform, benefiting from all the various features that will be implemented in the meantime.
  2. project sale of the phase of:
  • customization of the AI Assistant Platform based on specific customer needs (e.g. custom graphics, integrations, etc.)
  • identification, proposal and choice of the AI engine with the customer
  • implementation of integration with the AI engine
  • fine tuning (i.e. input/prompt refinement/integration/…)
Final notes: to ensure full mastery of the data by the customer, the AI Assistant Platform is installed on a cloud infrastructure owned by the customer or on a third-party cloud infrastructure (e.g. Azure, AWS, etc.) contracted and managed directly by the customer.

IDE: Visual Studio Code
Technologies: Docker, MongoDb, Expressjs, TypeScript, JavaScript, HTML5, CSS3, Angular, Bootstrap, Node.js, DevExtreme, OpenAI REST API, Gpt4 REST API
Share facebook share twitter share pinterest share
Sigla.com - Internet Partner