Authored by Sameer Dhanrajani, Chief Strategy Officer, Fractal Analytics
In my previous blog, I touched upon the evolution of the data-driven decision-making and how Artificial Intelligence (AI) is disrupting the Business Intelligence (BI) arena. The massive scale of data detonation seen across global enterprises today combined with the need to take appropriate business decisions in real-time has created a massive need for AI to augment current BI systems. Existing business intelligence systems served great value in the past, but the dynamic landscape of business today requires a different, much faster response and AI is most certainly going to be that game-changer. This often leads to the question – what happens to BI then? Are we slowly seeing a sunset for this technology? The answer is – absolutely not. Business Intelligence will continue to serve a key part of the data-driven business decision-making paradigm in the AI era. What will happen however, is that there will be an incumbent expectation that existing BI tools, processes, systems and people be heavily augmented by Artificial Intelligence. AI will bring out the best in BI – by unlocking an even higher business value of data and driving higher adoption and activation.
AI Augmenting BI for Exponential Business Value
Let us look at some flavors of AI that will profoundly impact how we use business intelligence and drive exponential value from our enterprise data.
Computer Vision and NLP to Unshackle Unstructured Data
One of the key challenges with vanilla BI tools is their underwhelming capability to handle unstructured data such as voice, text, image and video. While we have cracked the code for the capture, storage and retrieval of such unstructured data through sophisticated data lakes and warehouses, insight generation is still heavily reliant on manual intervention and is extremely time consuming.
By incorporating AI techniques such as Computer Vision, we can automate insight generation and extraction from image and video files. For instance, we are seeing numerous examples in this arena where Computer Vision driven solutions are identifying and quantifying data from image and video files – from quantifying car damage for speeding up car insurance claims to identifying anomalous regions in CT-Scans. Similarly in Natural Language Processing, there are exciting use cases of sentiment extraction from customer service call and email logs. While we’ve had the capability to store and index this information adequately well, AI is helping take the analysis of unstructured data to the very next level.
Voice and Text Based Search to Democratize BI Access
One of the key challenges faced by users of BI today is its siloed nature. Business Intelligence software is often very heavily cordoned off, with access available largely through technical administrator teams. Knowledge of the relevant querying language becomes essential to extract any meaningful information from these systems and generating custom reports is often a cumbersome and time-consuming process, leading to lost time and possible revenue opportunities.
We can drive much more meaningful business value from these BI systems by overlaying an element of natural-language driven search. With advances made in natural language understanding and natural language generation, business users can pull out ad-hoc data that is often needed on war-footing and cannot wait for someone with knowledge of ‘data-speak’. Conversational and Cognitive BI have often been touted as the panacea for these problems and such systems are rapidly hitting the mainstream. One such search-enabled BI platform is provided by ThoughtSpot – a startup that recently hit the much vaunted unicorn status. Artificial Intelligence techniques are a huge boon for senior business executives – enabling increased democratization of access to data and putting them in the driver’s seat for making highly accurate strategic decisions.
Machine Learning for Extracting Real-Time Value from Big Data
We already discussed the data detonation that we are seeing today. With increased sensorization and systematization, we are generating more data than ever before, with high velocity and complexity. Making sense of this high-speed, yet critical data is often beyond the human realm. A machine-augmentation of human capability is then of immense value to extract insights from such huge and multi-source data.
This issue can be effectively handled by Machine Learning (ML) driven systems. ML techniques are well geared to parse through very large data sets streaming at high speed to extract very relevant and crucial insights. What’s more – Machine Learning algorithms are geared to continuously learn from the data at their disposal and continuously furnish more relevant and accurate data.
Real Time Anomaly Detection
One very key issues that data teams are often tasked with is identifying anomalies in a largely monotonous data set. However, with the data deluge we are seeing today combined with the extremely siloed nature of current BI systems, it is increasingly harder to comb through this data to find insights that could have massive bottom-line impact.
Instead of throwing more people at a massive problem, AI-powered algorithms are increasingly adept at ingesting large data to spot and explore anomalous conditions and report their causes. For instance, ML algorithms can very quickly go through sales data at a brand and product level – to identify if there was a trend reversal or a non-occurrence in retail sales for a CPG organization. The ability to know and diagnose such instances early can help brand managers make hugely important decisions early and take the required corrective action to stem possible revenue leakages early. An Israeli startup – Anodot – has been making impactful strides in the space of real-time anomaly detection to identify outlier conditions in time-series data.
Chat-bots as Business and Data Analysts
If the existing paradigm of Big Data is beyond the human realm, what can be said of data and business analysts that currently are engaged in deciphering insights from data? Will AI change the way in which data is consumed by organizations?
We need to reimagine the human-machine equation in the field of data and insight discovery. Already we are seeing very exciting work by some BI companies that are increasingly turning to chat-based systems allowing users to ‘talk’ to a virtual business analyst to make sense of their data. Chat-bots today are fairly capable of understanding questions asked in a natural language, understanding the preferences of the user – to furnish personalized data and reports. Sisense is an example of a company doing breakthrough work in this space. Sisense Boto goes a step further by integrating their bot in your preferred messaging applications as well – Slack, Skype and Facebook Messenger – to make the data even more accessible for business users.
As we talk about an AI transformation and disruption in the field of BI, we also need to consider which approaches work best for individual organizations. There are multiple ways to instrument this transformation – from building an in-house AIML capability, to sourcing increasingly AI-centric BI systems and augmenting capabilities from external specialists. Changing the paradigm of BI in the AI era will require a robust strategy - encompassing audit of existing tools and systems, reinventing the data culture of the organization and migrating the in-house talent to build and manage such critical systems. By incorporating these fundamental constructs, we will be able to bring to life the new-age of BI, one that is powered by Artificial Intelligence and can capably unlock more value from your data. The future of BI is AI. Are you ready for the transformation?
No comments :
Post a Comment