All Collections
AI tools
Labguru Assistant - How to Select Powering Model
Labguru Assistant - How to Select Powering Model

Explore the capabilities of different AI models within Labguru Assistant and learn how to choose the one that suits your needs.

Eynav Haltzi avatar
Written by Eynav Haltzi
Updated this week

Introduction

Labguru Assistant is an AI-powered chat interface that seamlessly integrates with your Labguru workspace, providing users with immediate assistance and support directly within the application
The Assistant supports scientists and researchers in their daily tasks. It can act as a virtual lab assistant, offering immediate, AI-driven responses and actions based on user queries and commands.

Powering Models and Data Privacy

​Due to the critical importance of data privacy and security, Labguru offers two language models to power the Assistant. The selection of which model is used in a Labguru account is available only to the account owner from the account's setting page.

The models available for selection differ in their source and where the model is hosted and therefore their data privacy considerations are different. In addition, it's important to note that each model has different capabilities and may perform differently based on the specific use case. The model selection and data privacy settings are adjustable and can be altered (by the account owner only) at any given moment.

Note: Regardless of selected model, users should always review Labguru Assistant's responses, as they may occasionally be inaccurate or incorrect.

  1. Labguru in-house model

    The Labguru in-house model is hosted on Labguru's servers. The data remains confined to Labguru's servers and isn't shared with external third-party services. There is no selection of data privacy level in this model.

  2. OpenAI API platform

    The Labguru assistant connects with Open AI's API for text generation tasks. OpenAI are committed to privacy and security for their API platform as detailed in the following article.

    OpenAI services are also subjected to Privacy Policy and Usage policies and under DPA (Data Protection Agreement) with Labguru. In addition, OpenAI complies with several security-related regulations detailed in the security portal.



    In the case of Labguru Assistant powered by OpenAI, there are four data privacy levels available for the account owner to select from:

    1. Full (Account database) - Labguru Assistant will have access to the whole account database and will have the ability to reply to user queries with direct relation to the account data. Assistant responses are still subjected to the user access level and will not reveal to the user any information that s/he is not authorised to view.

    2. Specific page Integration - Labguru Assistant will have access to the data available on the currently opened page while a user performs a query. As a result, the Assistant will have the ability to reply to users' queries with direct relation to that specific page they have opened while submitting a question.

    3. User preference - In this mode, the Assistant will ask each user to define personal data-sharing preference. Users will be able to select between the option to share specific page data (option b) or to work in prompt mode (option d).

    4. Prompt mode - Labguru Assistant operates without access to the database. Instead, it relies solely on the text actively inserted into the chat text field by the user during a conversation

Important: In both models, each Labguru account has its own Assistant instance, and it's data is not shared with other Labguru databases (of other customers).

Model-Based Functionality

Action

Labguru Model

OpenAI

Answer specific page related questions

Yes

Yes

Answer cross account context questions

TBA

Yes (based on the global search)

Answer index page related questions

Yes

Yes

Create a document

Yes

Yes

Search PubMed

TBA

Yes

Search the Help Desk

TBA

Yes

Investigate datasets

TBA

TBA

Did this answer your question?