Transforming Due Diligence with AI: From 3 Months to 3 Weeks

Tags: AI analytics, due diligence, data transformation, investment, chain-of-thought reasoning

The Challenge: AI transformation in data analytics

Our client, an Investment agency with heavy due diligence needs, approached us as they were looking to move away from their existing analytics services provider: the lengthy timelines of up to 3 months to complete data analysis and visualisation tasks, paired with a cost of $10k per company analysed were not efficient and sustainable enough to allow them to scale. They were also looking to hold the entire due diligence process in-house while reducing time and cost.

Our Solution: Custom AI analytics

To help this investment agency, we developed a custom AI analytics solution to significantly reduce time spent on due diligence processes.

Here's how:

• Data Ingestion: From 1 Month, down to 1 Week — The first step in any due diligence process is gathering data from various sources. Previously, this process took up to a month, as the agency manually collected large amounts of data from many different sources, such as sales numbers, advertising results, and operations metrics. The first capability of our AI Analytics Tool is to automate the data ingestion process: pulling data from various sources, cleaning it, and sorting it for analysis. This reduces the time needed for data ingestion from one month to just one week.
• Data Transformation: From 2 Days to 5 Hours — Once all the data is gathered, it needs to be aggregated and transformed into a format suitable for analysis. Previously, this step involved writing SQL scripts that had to be adapted to different data sets — a process that took up to two days for each project. Our AI Analytics solution automatically adapts SQL scripts based on the data set it's working with, reducing the time required for data transformation from two days to just five hours.
• Analysis & Visualisation: From 2 Months to Real-Time Insights — The most time-consuming part of the due diligence process is the analysis and visualisation of the data. The agency worked with PowerPoint presentations of up to 500 slides. We changed this approach by developing an AI-driven analytics interface where users could ask specific due diligence related questions, such as "Which country should we invest more in over the next 12 months?" The LLM deeply understands the database it has access to, allowing it to provide data-validated answers, as well as chart & table options for visualisation.

The Results: An absolute game-changer for Due Diligence & Data Analytics

Our AI Analytics Tool takes care of the process end-to-end, from gathering and cleaning data to analysing it to answer questions. Not only does it translate into considerable time and cost savings, but it also means better and faster decision-making.

Key Outcomes:

• 75% reduction in time spent on data gathering after implementation of AI Analytics
• 77% reduction in time spent for all due diligence data analysis processes: from 3 months to under 3 weeks
• Real-time data analysis and chart creation for all types of investment and data-related questions

How does it work? The Tech Note

Linked to our AI Analytics tool is a customised LLM that can answer complex data analytics questions.

To achieve this, we use Chain-of-thought Reasoning: where the AI automatically creates a set of prompts for itself after being asked a question by a user, allowing it to provide high-quality answers — the same way we as humans think step by step when problem-solving!

Let's take an example: say an analyst asks the AI Analytics tool "Which country should we invest more in in the next 12 months?"

To answer that question accurately, our AI will first generate a list of 10 to 20 questions for itself, such as "Which country has the most revenue?" or "Which country has the highest marketing spend?". For each of these questions, the AI then creates a query, runs it to pull data from the database, and creates the corresponding graphs or charts. It then uses all that information to make a conclusion.

Pairing an AI Analytics Tool with a customised LLM not only saves time but also solves the problem of accessibility of data analytics. Virtually anyone can use the tool to make data-informed decisions, without needing any data analytics knowledge.