Docs

Docs

Language Models for Analytics, Agents as Proxies, our Vision

Language Models for Analytics, Agents as Proxies, our Vision

Language Models for Analytics, Agents as Proxies, our Vision

(Click to view video)

What is Patterns?

Patterns is a self-serve analytics platform revolutionizing how data is utilized in organizations. By integrating AI-driven agents, Patterns allows business users to access, analyze, and interpret data without extensive technical expertise.

Patterns is designed as an analytics platform powered by AI agents, which act as proxies for real people within your company. These agents enhance intelligence and impact by automating data-related tasks without the need for incremental human effort.

Key Features

- Database Integration: Agents seamlessly integrate with your data warehouse and provide intelligence out-of-the-box.

- Agents as Proxies: Agents function as an extension of your data team. By connecting your data warehouse and knowledge base to a language model, providing a sophisticated interface for data interaction.

How Agents Work

Agents transform user queries into actionable insights through a multi-step process:

1. Request Handling: Upon receiving a user request, agents create an action plan.

2. Data Research: They assess all relevant data tables to gather necessary information.

3. SQL Generation: Agents generate SQL queries to retrieve data. If errors occur, they automatically correct these queries.

4. Chart Creation: Alongside data retrieval, agents create visual representations of data. Any inaccuracies in charts are self-corrected.

5. Insight Generation: Agents analyze the collected data and charts to generate comprehensive insights, which are then formatted into reports for business users.

The Future of Patterns

Looking ahead, the role of agents within Patterns is set to expand significantly:

- Proactive Insights: Agents will proactively deliver insights, querying data and suggesting improvements to the data model.

- Actionable Analytics: Beyond analysis, agents will execute data-driven actions, such as compiling lists of new users and initiating contact through automated emails.