AI Business is part of the Informa Tech Division of Informa PLC
This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 3099067.
LONDON, UK - Cracking real-world problems with AI, whether they're B2B or B2C challenges, calls for a business-first, human-led approach to AI implementation. By looking at every client relationship through the eyes of our clients, the customers, and the market in general, you can gain a distinct advantage - and for us, it means the bespoke software we offer our clients can really drive transformative results.
One of the first challenges AI vendors face is understanding the client's business and customer base. Ideally, the AI vendor should have real world experience in their customer's space, as well as an understanding of their client's customer profiles and motivations. AI is bespoke and all clients are different, each with their own idiosyncrasies and USPs. Back office theory can only take a vendor so far, and data analytics can take them even further. However, it's ultimately experience of the real world of business and the opportunities it presents that will cut out a lot of hypothesis and guesswork.
Then there are integration challenges. A banking client had recently purchased an automated machine learning platform (AMP), but was unable to fully utilize the platform. This was due to a lack of inhouse system integration skills, as well as the fact that the AMP system itself was trained on US datasets. AI SENSUM approached the bank to help integrate the AMP system with their legacy fraud detection and churn management systems, with minimal day-to-day disruption and higher efficiency returns.
Any AI system is dependent on two things: the quality of the training dataset and a feedback loop mechanism for continuous improvements. We have deployed a sales rep product portfolio optimization software to a leading beverages company with field feedback-enabled features. This allows sales reps to put in qualitative and quantitative feedback based on their field trips, which is then analysed to make better predictions for any future visits. The AI system prediction relevance improved significantly following the launch of these field feedback features.
An AI solution is hugely dependent on the quality and abundance of the training dataset - especially in deep learning scenarios where many training data points are required for predictions.
The organizations should look at their data as a revenue generator, rather than as a cost. Investing in better data management practices, as well as gathering market intelligence and customer feedback, can ultimately produce better AI solutions to aid in increasing sales and reducing OpEx.
Skillsets also remain an issue. Today's organizations already tend to have data mining experts, or at least some level of predictive analytics within their insights and marketing practice. They should encourage their teams to understand the concept level design of existing or potential AI solutions. Inhouse experts can help track the performance of AI solutions over legacy processes and be the flagbearer of AI practices within the organization.
More than anything else, culture remains a challenge. Enterprises should work with vendors with broad market experience; not just a vendor that has outgrown their space, but a vendor with an objective, helicopter view of how to make things work for their customers.
For project preparation, the client needs to get the AI vendor out of the data and into the boardroom. At the very least, an in-depth workshop involving all stakeholders is a good starting point. Many AI companies focus only on the algorithms and their product toolkit, even though cultural understanding and a consumer-centric mindset is an integral component of understanding and implement an effective AI solution. By openly sharing our source code with clients, we ensure multiple stakeholders engage with and even champion non-black box AI within their departments and across the business as a whole.
Many AI providers, including AI SENSUM, are pursuing a modular solution approach to help tackle a client's business problem. These AI modules are designed to sit within legacy systems and help improve the system performance over time. These are low-cost bets for companies who are unwilling to invest in fully automated AI platforms.
Organizations who adopt AI early on will have a massive advantage in terms of efficiency gains. Late adopters to AI will be unable to bridge this lead, and are unlikely to succeed in the next five years. For new adopters of AI, we propose a data discovery module, in which we ask for a slice of data from the client for analytics. A proof of concept (PoC) piece of software is created which answers some of their business questions. The clients are then able to see a real-time, working prototype using their own data which then delviers belief and trust in the AI system we develop. Once the client agrees on the PoC, we then go AI-LIVE, which integrates the system permanently into their IT infrastructure.
That's not to say AI has all the answers. People need to sit alongside AI platforms to truly make them a success. AI technologies, coupled with market and business sensibilities, is essential.
In any effective AI solution, there must be a way for humans to sense-check, provide qualitative feedback, and monitor improvements over time. An AI solution without human-driven business intelligence can even take longer to produce results than existing systems.
Our model of customer centricity - of placing the consumer cultural understanding before the algorithm - is what we believe will make our implementations standout from the algo preachers. We can do this because our leadership team is global, and each member has a background in market intelligence and insight across borders and verticals, as consultants or clients.
Matthew Coulter has 15 years experience in Management Consultancy and delivering insights to clients globally across a range of verticals. He now heads up client relationships in Europe for AI SENSUM out of London. Matthew brings real-world understanding to clients and coordinates a world class back office team to ensure that clients monetize their data and have actionable insights from their data.