Unleashing the Power of GPT: From AI Assistants for Analytics to our own J.A.R.V.I.S.?

Introduction

In 2013, we embarked on an ambitious journey with the vision to create an intelligent agent to assist business users in data analysis. Initially, we started with natural language search, progressively moving towards automation of analysis, incorporating data science, and utilizing machine learning. This evolution included generating narratives that provide context to the data, aiding business users in understanding the intricacies involved.

The introduction of Generative Pre-trained Transformers (GPT) has significantly enhanced our capabilities. The technology has made it easier to interpret user inquiries and deliver swift solutions. Additionally, GPT has enriched the narratives we generate, making them more meaningful and comprehensible for business users.

One of the fascinating insights we've gained is the potential to harness the vast volumes of unstructured data that businesses possess. It’s estimated that only 20% of a company's data is in structured databases, while a staggering 80% resides in unstructured sources such as emails, PDFs, and various text-oriented documents. Large language models like GPT excel in understanding and analyzing this unstructured data, which allows us to broaden our analytical capabilities significantly.

Answer Rocket has the unique ability to aggregate relevant data to produce answers that align with the queries posed by business users. GPT serves as a vital bridging element, enhancing our understanding of user questions while facilitating a better comprehension of Answer Rocket's capabilities by the users. This synergy is pivotal in transforming traditional dashboard retrieval methods into an engaging conversational experience, effectively turning static reports into dynamic interactions with an intelligent agent.

Our ultimate vision for AI assistants is to create a collaborative environment where the AI feels like a team member. We envision a world where employees can delegate tasks such as research, summarization, conclusion drawing, and even presentation building to AI agents. This evolution would enable every human employee to function like an executive, managing a team of AI agents tasked with various responsibilities.

As AI models expand in size and sophistication—like the advancements seen with GPT-4—they become increasingly adept at processing larger data sets and approximating human decision-making. With ongoing training on key business concepts, these models will evolve from sources of basic information to offering genuine expertise and partnership to business executives.

If we reflect on popular culture, Tony Stark's J.A.R.V.I.S. serves as an inspiring example of a sophisticated assistant. Just a year ago, I would have estimated that it would take about 20 years to reach a level similar to J.A.R.V.I.S. Today, however, that timeline feels significantly shorter. We are, indeed, getting closer to realizing this level of intelligent assistance capable of guiding users, whether they are overseeing digital advertising or managing complex operational aspects of a business.

Keywords

• GPT
• AI Assistants
• Data Analysis
• Unstructured Data
• Intelligent Agents
• Collaboration
• Business Insights
• J.A.R.V.I.S.
• Automation

FAQ

Q1: What was the initial goal when starting the company?
A1: The initial goal was to create an intelligent agent that helps business users with data analysis.

Q2: How has GPT enhanced the user experience?
A2: GPT has made it easier to understand user inquiries and deliver more meaningful narratives, enriching the user experience.

Q3: What percentage of a company's data is structured vs. unstructured?
A3: Approximately 20% of a company's data is in structured databases, while 80% is unstructured.

Q4: What is the ultimate vision for AI assistants?
A4: The ultimate vision is for AI to act as a collaborator, allowing employees to delegate tasks and effectively manage a team of AI agents.

Q5: How close are we to having AI assistants like J.A.R.V.I.S.?
A5: With recent advancements in AI technology, we feel we are much closer to achieving the capabilities similar to those of J.A.R.V.I.S. than previously estimated.