Towards the Universal Enterprise AI
Evo achieved Quantum Learning for better Enterprise Artificial Intelligence
The Enterprise Learning Challenge
In spite of the hype, companies are still grappling with the challenges of integrating Artificial Intelligence (AI) into their decision-making processes. Concerns about data quality, biases in AI algorithms, lack of talent, and ethical implications of automated decision-making pose obstacles, limit adoption, and reduce the potential efficiency and accuracy. Balancing the benefits of AI with its challenges presents ongoing struggles for companies seeking to harness the full potential of AI in their decision-making processes.
Driving personalized learning at the granular level of each customer and/or product is particularly daunting for companies adopting AI in their decision-making processes. Tailoring AI algorithms to individual customers, products and locations requires vast amounts of high-quality data, sophisticated machine learning models, and constant refinement. Companies must also consider privacy concerns, data security, and customer trust while striving to deliver truly personalized experiences. Achieving this level of granularity in enterprise learning poses significant challenges that companies must navigate to fully benefit from AI in their decision-making processes.
Therefore, it is not surprising that such challenges make companies cautious in embracing AI: managers still often rely on traditional analytical methods instead of fully adopting Artificial Intelligence (AI) for their decision support. Traditional analytical methods may seem more familiar and manageable, and companies may prefer to mitigate risks by using established approaches, despite the potential benefits of AI in decision-making processes.
But such traditional approaches may cause, among other outcomes:
- Low profit margins
- Missed sales opportunities
- Waste of product inventory
- Loss of customers
- Inefficient operations
Enter Quantum Learning
“Quantum learning” has been used to describe learning techniques based on principles from quantum mechanics, a branch of physics that deals with the behavior of particles at the atomic and subatomic level (as by defined by AI via GPT).
In the same way that quantum computing allows you to process more information incredibly quickly, Quantum Learning delivers that same speed without sacrificing complexity. With a quantum approach to learning, you can address all these factors at once just as quickly as traditional learning methods make extracting more basic insights.
At its core, Quantum Learning involves the idea that learning is a complex process that involves both the learner and the environment in which the learning takes place.
According to this perspective, learning is not a linear, cause-and-effect process, but rather a dynamic, non-linear process that is influenced by a variety of factors.
In a similar way, enterprise learning is a dynamic process that involves both the customers and the managers’ environment in which the learning takes place. It is therefore a reflexive challenge, where the learner (manager) and learning objective (predicted customer behaviour) are not linked by a linear cause-and-effect process, but rather a dynamic non-linear process influenced by a variety of factors including the manager’s own vision and decisions. A perfect context for Quantum Learning: learning as much as you can about the customer within a massively complex environment.
The term “quantum learning” has however been criticized by some experts as being vague and lacking a clear definition or empirical basis. Some argue that the principles of quantum mechanics do not directly apply to enterprises: while the principles of quantum mechanics have been successfully applied in fields such as computing and cryptography, their direct applicability to enterprise learning challenges is still an area of active research and exploration.
Our own research has therefore narrowly defined Quantum Learning based on two critical characteristics:
- Granular level: the level of detail that can be divided to learn more is practically infinite. Even the single customer, for example, can be further differentiated based on the combination of product, price point, time, sales person, channel, and many other factors — a “quantum level” is therefore a level of learning at which traditional models fail (think “indetermination”)
- Universal type: lessons need to be applied across a practically infinite number of variables. At such granular level, the structural challenges of learning converge, whether for a B2B or B2C customer, for a perishable or non-perishable product, for a low or high price/promotion point, for a Chinese or Chilean location, for a human salesperson or for e-commerce — “quantum type” of learning therefore defies the gravity of traditional models (think “theory of everything”)
Our research also revealed additional interesting typically “quantum” characteristics on top of the two main ones (everything from entanglement to duality and state, which we leave for future posts). Ultimately, the parallels seem to ultimately run deeper and wider than some experts’ skepticism would otherwise suggest; once the Quantum Learning perspective is embraced, it can become liberating and tremendously effective.
Achieving Quantum Learning Now
Imagine; what if you could model each individual customer and/or inventory unit as the truly unique phenomenon that they really are?
Well, now that’s possible.
After many years of patient data collection and research, Evo recently announced to have “Achieved Quantum Learning in Enterprise AI Applications”, which was met with a mix of curiosity and skepticism.
To quote the media coverage:
Quantum Learning is the result of a predictive model of customer choice that breaks the Quantum point: both universal and granular at the lowest possible level. Until today, it was not possible to model each combination of customer, product, location, and price point using a single tool. Moreover, even the most sophisticated models required massive development lead times and additional data collection before being ready for production.
This achievement unlocks autonomous decision-making for business management without specific training or data collection relative to each company, bringing better management decisions through data to companies worldwide. EvoAI no longer requires massive data ingestion or long integrations. The autonomous data engine adapts immediately to any business reality and generates recommendations on pricing, inventory management, and customer scoring. The Quantum Learning model cuts waste, increasing supply chain sustainability, and improves profit margins. (…)
The AI software can be plugged into any company with transaction data. The software immediately begins learning and delivering impact within hours, not months. These innovative tools can inform a range of decisions, from the best contract to recommend to an enterprise client to the optimal inventory levels and prices for a single store to the right customer profile to maximize sales. Quantum learning represents the next level of customer-centricity and service granularity not previously attainable by other AI technologies.
EvoAI focuses within two main application areas:
- Supply-chain: the product-driven understanding that each combination of product, location and price/promotion is truly unique
- Customer: the customer-driven understanding that each of combination of customer, channel and price/promotion is truly unique
Towards the Universal Enterprise AI
The quantum learning milestone is the figurative summary for a number of additive accomplishments:
- Unified Data Model: abstracting the specific unique characteristics of each enterprise context. “Structured data” is surprisingly complex and unfriendly to the Artificial Intelligence models used for text, images and audio data, thus requiring a targeted healthy dose of basic scientific research; however, ultimately all transaction/order data is about “someone buying something” and as such it can be organized to allow for cross-learning that is context-free. Think “anything can be an attribute”.
- Data Variety and Velocity: achieving fast and relatively inexpensive results which go far beyond what would otherwise be possible by relying on “internal company data”, while complying with the challenges posed by regulation (e.g. GDPR) since Personally Identifiable Information (PII) is relatively uninteresting to Quantum Learning, which instead benefits from cross-fertilization. Think “anything can be relevant”.
- Specific Application and Measurement: demonstrating that the theoretical gains unlocked by Quantum Learning can and do translate into tangible benefits that can be captured by “the rest of us”, ordinary human beings who are as impatient, easily confused and skeptical just like you and me. Think “the proof is in the pudding”.
Above and beyond milestones, accomplishments and media coverage, this is still an area of active research & development. The Harvard Business School published two cases studies of relevant specific applications: AI-Driven Pricing & Promotions at Pittarosso and Merchandising at Miroglio Fashion — but there is so much more still to be discovered.
Vertical applications to address specific management challenges, and the reduction of friction in the adoption of Quantum Learning solutions, is equally if not more important than the underlying scientific and technical accomplishments.
One thing is certain: “quantum learning” does not necessarily require “quantum computing”, which is a specific infrastructure set that may be used to implement certain types of computer code.
In fact, the world’s largest quantum computer has 433 qubits. Much as we love to think of them as “cool bits”, they are still many years away from reality. Maybe they can maybe help with some specific optimization tasks, but a Quantum Learning enterprise model requires lots of other non-quantum computing infrastructure (hype aside). Quantum Learning is an approach towards an outcome, not a single technology.
Onwards and upwards
Making Artificial Intelligence work for every manager is an exciting challenge. So much more remains to be discovered, and the readiness of the AI and Enterprise communities is still low. After all:
- we have mostly not covered these topics during our academic studies
- there is plenty of confusing over-hype around the subject to confuse us
- the Quantum Learning technology transforms traditional boundaries of org charts
What if the supply chain was not measured as a back-end function anymore, but as the integral part of the customer promise which it arguably already is?
What if marketing was not measured as a specific set of activities anymore, but as the pillar of all steps across the end-to-end customer journey which it arguably already is?
Quantum Learning makes all this possible. It just takes a change in mindset.
If you are ready to transform not just what you know but also what is even possible to know about your customers, it’s time to take the Quantum Learning approach.
Every business leader today knows that a customer-centric strategy is critical to compete. Quantum Learning allows you to leap forward in your search for that vital knowledge.
Fascinating, thought-provoking ideas. Want to learn more, ask questions or comment? Feel free to reach out below, or via Linkedin, or through our website.
Our plan is to continue this discovery journey.
Onwards and upwards: towards a better AI that works for every (human) manager.
About Evo Pricing
Evo is a leading London and Turin-based tech company that uses AI to optimize supply chain, pricing, and customer decisions. Since 2015, Evo has helped clients leverage Big Data to increase the customer relevance of business decisions, which helps reduce waste, optimize market efficiency, enhance product availability, increase margin and raise service levels. Evo’s algorithm started as Evo CEO Fabrizio Fantini’s PhD thesis, as a way to address limits of traditional pricing software. Since then, Evo’s applications have increased in sophistication and accuracy, as well as scope, with autonomous solutions covering price management, markdowns, forecasting, customer scoring and replenishment.
About the author
Fabrizio Fantini is the brain behind Evo. His 2009 PhD in Applied Mathematics, proving how simple algorithms can outperform even the most expensive commercial airline pricing software, is the basis for the core scientific research behind our solutions. He holds an MBA from Harvard Business School and has previously worked for 10 years at McKinsey & Company.
He is thrilled to help clients create value and loves creating powerful but simple to use solutions. His ideal software has no user manual but enables users to stand on the shoulders of giants.