The easiest way to get GenAI into your CRM

Currently, Magnet AI can be deployed only to the Microsoft Azure cloud. In the coming months, we will make it available first on OCI and then on another cloud platform, depending on the customer's interests. 

We integrate in two ways. First, the CRM system can invoke Magnet AI API, providing unstructured or JSON-formatted text as input and receiving the output as JSON. Second, Magnet AI Panel can be integrated into the hosting CRM UI natively or as an iFrame. When it comes to Magnet AI's mini-apps, they can read and update CRM data using standard REST services.

Azure OpenAI never trains its standard models on the data you pass to it via API. At the same time, the request and response data are stored in Azure for a certain period, and Microsoft employees can review this data to assess potential service abuse. Customers may request Microsoft not to store this data to achieve ZDR (Zero Data Retention). Read more about Azure OpenAI service data, privacy and security here.

Absolutely! On this webpage, we try to avoid technical terms such as RAG (retrieval augmented generation) and refer to it as Q&A flow or semantic search. While the RAG concept is very straightforward, there are various techniques to improve the results, and we are gradually implementing them in Magnet AI.

It depends on your use case. Assuming that you can set up Azure resources and provide access to your knowledge bases and CRM system API without delay, it will take just a couple of days to start testing RAG and prompt templates' API. Implementing and testing custom mini-apps might take from several days to several weeks.

For prompt completion, we use Azure OpenAI GPT-3.5 Turbo, GPT-4, and GPT-4 Turbo, and for embeddings, Azure OpenAI Ada-002. As a next step, we will start leveraging Cohere models due to their performance and availability both on Azure and OCI. LLM space is very dynamic, and we keep monitoring it, paying close attention to Llama, Mistral, and Claude models.

We have built Magnet AI Admin UI so that your current CRM administrator or maintenance team can operate it while implementing an increasing number of use cases. While Magnet AI's defects can be reported and most likely will be solved, the MIT licence does not provide any SLA. However, you can request premium support services from Magnet AI implementation partners.