LinkedIn

Twitter

GitHub

Whitepaper

<h2>Introduction</h2>
Launched by OpenAPI at the end of November, ChatGPT has changed the perception of what’s possible using Artificial Intelligence. A wide range of industries and use cases are adopting ChatGPT for its ability to reduce manual effort and accelerate operations. Not only can ChatGPT summarize information, synthesize new content, and assist development, it can also supercharge your analytics. In this white paper, you will learn how to use it to save time and effort in your operational analytics processes and gain more insight.
Built on top of GPT–3.5 as a Generative Pre-training Transformer language model, ChatGPT responds to a broad range of user inputs. As a result, ChatGPT can support a massive array of use cases, from automating coding, creating customer service responses, providing language translation, and more. It’s even poised to change how we regularly get answers to simple questions that we currently use Google for today. 
ChatGPT is starting to change the way we work, but it is already clear that AI models can reduce costly operational overhead. For example, organizations routinely collect a wide range of ‘write-once, read-rarely’ profile information, such as company size, industry segmentation, or funding details. This type of information is generally available from multiple sources and only valuable when it’s needed. Plus, It usually requires significant curation and reconciliation.
Automating enrichment and augmentation of analytics with new fields, segmentations, and classifications is a challenge that analytics professionals and operations face daily. They must choose between spending costly hours grooming information that will likely never be used or waiting until it is required, when the lead time to gather it from external sources is prohibitive. 
By operating as an on-demand source for augmented and ancillary information, ChatGPT has the potential to eliminate much of the pain and effort associated with traditional data preparation and, in the process, enable teams to segment and slice data in new ways, on-the-fly.
However, applying an on-demand model is only feasible if ChatGPT is tightly integrated with your data and analytics tooling. Without it, the ease of self-service analytics is quickly lost. Seamlessly integrating ChatGPT to supercharge your analytics is now possible. It’s easy to do, and we think you’ll find it’s a game-changer.
<h2>Why you need to integrate ChatGPT with your analytics tooling</h2>
Teams often spend way too much time in manual data prep and enrichment today. Estimates range as high as business and data analysts spending 70 to 80% of their time getting data ready for analytics, a large portion of that spent enriching and labeling data and creating new attributes to aid analysis. Adding new fields to the data sets for analysts to slice and dice by is often painful, onerous work. For example, consider a seemingly simple query:
<blockquote>
What’s the breakout of my customers by industry?
</blockquote>
It’s the kind of question critical to making decisions about entering new markets and launching new products. But what if the thousands of customers in the data set were not segmented by industry beforehand? What if Industry isn’t a field in the data or data model? As an analyst, it almost certainly usually means you’re out of luck without investing significant time and effort.
It typically means going back to the data, adding external classification information to the data set, and reloading it in your data warehouse, and that’s incredibly time-consuming. In many cases, it requires manually matching with and reconciling third-party data sets, repetitive Google searches to tag customer records manually, and often painful convoluted IF () and VLOOKUP () formulae in Excel or custom Python script. Not to mention updating your data model. 
Worse, if teams subsequently need to classify and label data in a new way – such as by zip code or county, then it’s back to manually prepping the data again.
<h2>Gains from a tight integration of ChatGPT with your analytics tooling</h2>
ChatGPT is exceptionally well-suited for this kind of work, and it can do this kind of search, extraction and reconciliation automatically and quickly. However, selecting the right moment to weave ChatGPT-sourced information into your analytics experience is critical to make it truly useful. For example, collect this information during ETL, and both storage and query costs will stay high; collect it once requested and you may struggle to insert it into the data model to enable the user&#8217;s expected BI functions.
The optimal approach collects the requested data on-demand and uses ChatGPT to recommend the ideal insertion point in a model, providing a highly-responsive experience for the user. With this approach, users receive their own AI-powered always-on personal data prep and data analyst bot as part of their analytics experience. They can extend and augment their data and data model by simply asking ChatGPT. 
Achieving it creates a massive opportunity for teams, freeing them from manual work prepping and augmenting data while ChatGPT does the lifting, to spend time on more value-added and strategic work. But more importantly, beyond freeing up time, there’s transformational value to be had―enabling teams to be much more fluid in their analytical queries, segmenting data in new ways, with ChatGPT doing the work behind the scenes. 
<h2>Sisense + GPT-3: A new way to supercharge your analytics</h2>
A reference implementation of this strategy is available for Sisense Fusion, and is a perfect example of how to incorporate GPT models into analytics processing. This implementation leverages the uniquely flexible capabilities of Sisense Fusion to inject dynamic queries to GPT-3 as part of the query process, automatically generating new database tables and relationships in the existing data model while leveraging guidance from GPT-3. 
This reference architecture also demonstrates low-code app experience development, creating an interactive Q&amp;A app via Sisense BloX so users can tell GPT-3 how they want to augment the data set and model, and do so directly in their dashboards. Sisense and GPT-3 immediately handle the rest. 
How it all works can be demonstrated by answering the question posed earlier: “What’s the breakout of my customers by industry?”—when the customer attribute industry doesn’t appear in the original data model or data set. Sisense and GPT-3 will be used to fill the gap.
Below is the example data with a list of accounts. Ultimately, we’re looking for the percentage split of customers by industry even though the attribute doesn’t exist in the data.
<img loading="lazy" decoding="async" class="aligncenter wp-image-741530 size-large" src="https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-4-1024x920.jpg" alt="" width="1024" height="920" srcset="https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-4-1024x920.jpg 1024w, https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-4-300x269.jpg 300w, https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-4.jpg 1275w" sizes="(max-width: 1024px) 100vw, 1024px" />
To make it easy for the user to ask questions of GPT-3, Sisense BloX is used to build a simple interactive widget on a Sisense dashboard. This new UI assists in capturing what the user wants to request from GPT-3:
<img loading="lazy" decoding="async" class="aligncenter size-large wp-image-741528" src="https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-2-1024x352.jpg" alt="" width="1024" height="352" srcset="https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-2-1024x352.jpg 1024w, https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-2-300x103.jpg 300w, https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-2.jpg 1905w" sizes="(max-width: 1024px) 100vw, 1024px" />
In this example, GPT-3 is told the user wants the industry segmentation for a list of companies in natural language. GPT-3 is given the dashboard, widget, and column name, which ultimately contains the list of companies that need industry segment labels. 
The UI executes a Python script containing the question and the dynamically populated list of customer accounts formatted as JSON and submits it to GPT-3, which retrieves the industry for each customer.  Alternatively, if we want to add GPT-3 enrichment to the scheduled periodic data model refresh, simply executing Python rather than using the Q&amp;A UI, we can do that too. 
The result set is passed back from GPT-3 to Sisense Fusion as a JSON document. The Python script dynamically creates new tables in the origin database based on the GPT-3 result set and adds new relationships between tables in the model. Ultimately, it transforms hours or days of data enrichment effort into seconds.
In the example, GPT-3 returns each customer and their associated industry label in this example in JSON. Then, the script appends this as a new dimension to our data model and automatically joins our fact table of wins and losses with a natural join based on the account name. 
<img loading="lazy" decoding="async" class="aligncenter size-large wp-image-741529" src="https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-3-1024x833.jpg" alt="" width="1024" height="833" srcset="https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-3-1024x833.jpg 1024w, https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-3-300x244.jpg 300w, https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-3.jpg 1313w" sizes="(max-width: 1024px) 100vw, 1024px" />
With the augmented data added to our data model, it’s simple to continue our analysis. For example, we can quickly create two new visualizations using Sisense Fusion that show us the distribution of our customers by industry.
<img loading="lazy" decoding="async" class="aligncenter size-full wp-image-741527" src="https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-1.png" alt="" width="859" height="353" srcset="https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-1.png 859w, https://cdn.sisense.com/wp-content/uploads/chatgpt-whitepaper-1-300x123.png 300w" sizes="(max-width: 859px) 100vw, 859px" />
The whole integration enables the analysis to be completed in a few seconds rather than waiting days or weeks for an operations team to prep and publish the new information. Using the interactive BloX widget, augmenting and enriching any other data, all using natural language, is simple when powered by GPT-3.
As an exciting and related use case, a software product can easily add GPT-3 supercharged analytics by embedding this BloX widget directly into the user interface. As a result, users gain flexibility that extends the value of the product and its data by enabling augmented or ancillary data to be added without any involvement of the product vendor.
This approach to on-demand data isn’t necessarily easy to implement in a business intelligence product. The Python compute kernel and the extensive API surface in Sisense Fusion allow direct query creation and dynamic updates to data models to get this kind of on-demand user experience. However, just about any ETL engine can at least augment data at ingestion using this technique. Sisense’s reference architecture for GPT-3 gives you a good starting point to try these techniques out for your own use cases. Here’s a <a href="/blog/how-to-make-bi-and-analytics-adoption-simple/">link</a> to an embedded example of the exact analysis we shared. 
&nbsp;

How to Supercharge Your Analytics with ChatGPT AI