If you ask most companies about their data maturity, their response will usually cover one of the following phases:
Historically, there have been tons of solutions to address the first two of these phases. ETL and data warehouses to address Data Readiness, and no shortage of business intelligence and reporting tools to address Consumption Readiness.
The last phase, Realization Readiness, is typically the longest and hardest phase for companies to achieve. In part, many data and analytics projects end at the reporting or consumption layer from the onset. But a larger contributor is the difficulty of implementing downstream processes specific to your business. Since every business operates differently, this is the hardest part of the data maturity curve to create standardized solutions for. This means that most companies must rely on internal resources or partners to help build automations to power these processes. This is costly, time consuming, and usually deprioritized over other development needs.
Gen AI will supercharge the way you consume data. Through the use of generative AI, we can bring the realization of our data to the same place we’re consuming it. Shortening this gap allows us to achieve things faster and more efficiently than ever before. Let’s go through some example use cases to illustrate this.
Here are four typical buckets we see of use cases related to Gen AI for business intelligence:
Create. Gen AI’s best capability right now is its ability to, well, generate. Let’s say our Customer Service team wanted to send a personalized email to our most loyal customers this year.
Summarize. A popular use case for Gen AI is its ability to summarize or extract information, providing insights that may have otherwise not been uncovered. Perhaps our CEO has asked a business analyst to create a quarterly business review to see how the company is performing.
Discover. Datasets become even more powerful when enriched with other data points. Let’s see how we can improve the way we keep our CRM up to date and healthy.
Automate. The most sophisticated processes also require the most technical resources to implement. Automating these through the use of Gen AI creates endless possibilities for activating your data. Let’s look at a Customer Service team trying to improve customer experience and generate higher satisfaction scores for their website chat interface.
Before Gen AI, automating any of the processes above would have been long, complicated and cumbersome to build. As a result, most companies never recognize this apex point of the data maturity curve, where data doesn’t just inform, but directly impacts the way the business operates.
So why is Looker + Vertex AI the perfect pairing to solve this problem?
As our product team works hard to incorporate AI features into the Looker suite of products, Looker can already support your AI use cases.
Looker has always been a leader in extensibility. In other words, Looker’s API layer has always made it easy to send data from Looker to any downstream workflow or application. In fact, data scientists have been using Looker for years to power their AI/ML workflows by streamlining Data Readiness through its semantic data model, providing Realization Readiness through its RESTful API, and providing out of the box Consumption Readiness for analyzing results.
Looker’s Data Actions allows users to have a UI for inputting information or selection choices before triggering a task to be performed. This provides a mechanism for users to input a prompt, or a question, they want to ask of their data. It also allows for users to create schedules to trigger these requests. Schedules can be done on a regular cadence, or based on criteria thresholds.
By leveraging Cloud Functions, we’re able to send the underlying dataset to Vertex along with the user prompt. Furthermore, we’re able to send contextual prompts behind the scenes to fine-tune the model and produce better results. All of this is executed in your own organization’s Google Cloud project, safely and securely.
Diving more into Looker’s semantic layer, we’re also able to leverage Looker Blocks - pre-templated data models for common use cases. This provides us access to AI add-ons to our analytics.
Another piece of technology being leveraged here is BigQuery ML, which allows us to prompt Vertex AI directly via a SQL query in BigQuery using the new ML.GENERATE_TEXT function. Prompts occur at query time, at the row level. This is amazing for automating repeatable tasks for your organization.
Looker + BigQuery + Vertex is the all-in-one Gen AI tech stack that can help you improve the way data impacts your company. AI is your next best data analyst, ready to automate and augment data analysis tasks, scale your data analysis capabilities, and govern your data and AI models responsibly.
If you’re ready to leverage AI at your organization or learn more about Google Clouds' offerings, reach out to your account team to start the conversation. Have questions? Please leave a comment below. Thanks for reading!