Empowering Knowledge Workers with Vertex AI
The term “knowledge worker” was first coined by Peter Drucker in his book, The Landmarks of Tomorrow (1959). Drucker defined knowledge workers as high-level workers who analyse data and apply theoretical/analytical knowledge to develop insights, trends, products, and services. He noted that knowledge workers would be the most valuable assets of a 21st-century organisation because of their high level of productivity and creativity. “Organisations that rely solely on the IT department or analytics teams to fulfil queries around analytics are likely to be dissatisfied with the results”, says Alan Jacobson, Chief Data and Analytics Officer (CDAO) at data science and analytics firm Alteryx. Data is no longer a byproduct of transactional systems. Applications and technology need to be designed around an understanding of what data is needed to make better-informed, data-driven business decisions. The principle of silos or least privilege security model does not enable data-driven decision-making. Enterprises should make sure knowledge workers have access to the up-to-date and timely data they will need to run analytics, identify trends, and make informed business decisions. Organisations should make sure that knowledge workers have access to the best data tools [1].
In this article I will guide you through some of the features of Agent Builder, a new product recently introduced by Google in Google’s Next 2024. We will delve into its potential impact on the roles of supporting employees and knowledge workers.
A knowledge worker is a professional who generates value for the organisation with their expertise, critical thinking and interpersonal skills. Knowledge workers utilise data sources and various tools like data dictionaries, SQL query tools, Python packages, data visualisation tools, data manipulation tools such as PySpark, analytics, and potentially machine learning tools to derive insights, spot trends, and create products or services (Figure 1).
In modern workplaces, despite the abundance of data, integration issues often lead to fragmented workflows and hindered productivity. Employees frequently switch between tools and struggle with exporting and importing data, impacting their ability to effectively leverage data for informed decision-making. Timely access to large volumes of relevant data is also essential for accurate customer information and informed decision-making. Google Cloud’s Agent Builder and Vertex AI offers a solution to streamline data access and empower employees to make informed decisions and deliver exceptional customer service. Through this platform, various types of AI agents can be developed (Figure 2).
AI agents are the new hot craze in generative AI. They are similar to chatbots. However, unlike earlier chatbots, these agents are capable of more than just answering questions; they can act based on conversations and interact with backend datasources and transactional systems for automated actions.
Google Cloud CEO Thomas Kurian stated, “Vertex AI Agent Builder enables users to effortlessly and rapidly create conversational agents. You can develop and deploy fully operational, generative AI-driven conversational agents and train and direct them as you would humans, enhancing the accuracy and quality of responses from models.”
Overcoming Enterprise Challenges for achieving goals
Google Cloud endeavours to assist employees in achieving their most important goals through the Agent Builder by addressing the most important concerns of enterprise organisations:
1-Compliance and responsible AI: this entails preparing the enterprise to prevent data leakage, uphold data governance, safeguard data confidentiality, recognise data sensitivity, and oversee data usage.
This approach empowers organisations to maintain control over their data without utilising it for training or enhancing the quality of the models supporting such solutions.
2-Customisation and personification: customisation and personification are enhanced by integrating with enterprise systems through unified interfaces, which assist in managing tasks according to individual preferences and needs.
3-Factual accuracy: Improving the precision of AI responses by minimising errors and simplifying complex data to enhance factual accuracy.
Developer Options
The Agent Builder suite provides a comprehensive set of tools not only for building diverse agents but also for effortlessly integrating them into a unified system and a variety of messaging tools. This caters to users with different levels of data analytics expertise:
- In a low-code environment, agents can be quickly set up with minimal effort, personalised using NLP, and easily put into production (Figure 3). Quality assurance is supported through console simulators and testing capabilities before deployment to platforms like Google Chat or Slack.
- In a full-code environment, users have the flexibility to utilise their own code and Vertex AI capabilities, securely deploying applications on Vertex AI endpoints.
Grounding Gen AI outputs in Enterprise Data
Integration with enterprise data sources has been also made easy and secure, whether from cloud storage, BigQuery, public websites, or native integrations like Confluence or Hadoop file systems.
This integration ensures strict control over organisational data without utilising it for training or enhancing the quality of supporting models within the solution. Meanwhile, it empowers agents with vast volumes of internal data, enabling timely access to relevant data and improving efficiency and accuracy of provided services.
Developers have different options to use enterprise data for grounding:
- Vertex AI Search — Let Vertex AI Search streamline your workload effortlessly. Vertex AI Search is a component within Vertex AI’s broad suite of generative tools that excels in information retrieval and question answering tasks. This fully managed platform empowers developers to build Google-quality search experiences for their websites, structured and unstructured data. Easily integrated into any generative AI application leveraging enterprise data, Vertex AI Search serves as an out-of-the-box grounding system for the entire search and discovery process. From ETL, OCR, chunking, embedding, indexing, storing, input cleaning, schema adjustments, and information retrieval to summarization, Vertex AI Search streamlines these tasks into a few simple clicks. By basing LLM responses on your company’s data, Vertex AI Search ensures enhanced accuracy, reliability, and relevance — all essential for real-world business scenarios. Moreover, when utilising Vertex AI Search through Google Cloud, rest assured that your data remains securely stored within your cloud instance, safeguarded against unauthorised access or usage for training models or other purposes by Google.
- Retrieval Augmented Generation (RAG) Systems — A do-it-yourself option, leveraging Vertex AI APIs for annotation, document processing (e.g., complex PDF or .doc files), search capabilities, and ranking.
Vertex AI Agent Builder is also equipped with built-in security measures and enterprise controls, offering a comprehensive solution for swiftly developing and launching production-ready AI-powered experiences.
