Richard Potter, Co-Founder & CEO of Peak – Interview Series
Richard Potter is the Co-Founder & CEO of Peak, a platform that gives data engineers, data scientists and commercial decision makers everything to build and support AI-driven solutions across the enterprise.
Could you share the genesis story behind Peak?
The idea for Peak first started as a conversation at a pub on all of the different business intelligence products that were available at the time. My co-founders, Atul Sharma and David Leitch, and I wondered why so few companies could embrace data for decision making. We wanted a way to simplify things for businesses, to break down silos within enterprises so that teams could work together and everyone would be able to leverage useful outcomes based on data. This led us to the platform, which unites teams around a product built to optimize business with AI.
Could you describe what Decision Intelligence is for our audience?
Decision Intelligence is the application of AI to optimize commercial decisions. It is outcome focused, meaning DI solutions are built to deliver a tangible outcome, such as higher rate of sale or margin.
One of your predictions for going into 2022, is that a new discipline of data science is emerging. Could you elaborate on this?
As commercial investment in AI increases and data science matures, a new discipline of data science is emerging that starts with the end in mind.
Traditional data science projects begin by understanding the data available and what can be done with it. The result is hypothetical solutions to data problems, rather than AI solutions that can improve business performance.
By focusing on outcomes from the start of a project and understanding what is practical with the data available, this new discipline of data science prioritizes deploying solutions by starting with the end in mind. It enables businesses to get their AI deployed and unlock value from their AI strategy quicker.
Peak has built an artificial intelligence system that becomes a central system of intelligence within a company’s business. It aggregates data and deploys machine learning, to then output results. What types of machine learning algorithms are used?
The Peak platform uses a wide range of machine learning and modeling techniques – choice means we can tackle each project with the most appropriate method. We may use supervised and unsupervised methods, as well as forecasting or optimization techniques depending on the problem being solved. These can be built in our platform using Python, R and SQL.
With this flexibility and breadth of choice, Peak’s customers can build their own AI unique to their business. This is what organizations need to really embrace Decision Intelligence. Each company shouldnt have a standard AI, but something built specifically for them.
How does Peak enable companies to use their biggest asset – data – to increase sales and profits?
The Peak platform runs applications specifically designed to deliver on outcomes, be that increasing sales or growing profit (or both!). These applications span the world of marketing, sales, merchandising, inventory management, pricing and supply chain. Since it sits across an organization’s entire dataset, Peak’s Decision Intelligence platform can optimize across the whole value chain, providing real-time insights and recommendations that benefit every function within a business. This is a complex matrix and Decision Intelligence is the perfect tool for ensuring every decision made is right.
Peak is on first impression fully serviced, do companies using the service need to have AI engineers on board to use the platform?
The Peak platform has three core capabilities that allow users to:
- Combine data from across their organization and make it AI-ready.
- Build and train a centralized intelligence that uses AI models to provide a predictive view of their organization.
- Provide an interface for line of business users to interact with models that guide decision making across multiple functions.
Since it was founded in 2015, Peak has offered a model in which our platform and applications are implemented for our customers by our customer success and data science teams. We are increasingly seeing a growing number of Peak customers self-serving on the platform, building their own applications or deploying Peak's standard applications themselves.
What are some examples of Peak enabling businesses to optimize supply chains?
A good example would be a warehouse manager dealing with a stock issue. Traditionally they would need to manually bump up orders across overshopped SKUs, altering order volumes sporadically to account for volatility in demand.
But, with the help of a DI platform, the warehouse manager can be proactive rather than reactive. Taking into consideration circumstances across the business more broadly, the manager’s DI platform recommends that they decrease orders from the supplier. It sounds counterintuitive if there’s high demand, but the DI solution has identified that the company has a warehouse with a depot one county over with 2,000 units of that SKU that aren’t selling there. It’s already alerted the logistics team and has routed the scheduled delivery via that warehouse to pick up the additional SKUs. It will continue to run the same model to commercial teams across the business, adjusting the recommended action as data insights shift and each department takes action.
Another use case is reducing waste and energy, could you give some examples of clients achieving this by using Peak?
A global CPG retailer is currently leveraging Decision Intelligence to both optimize its transportation network and reduce the amount of wasted movements of goods between factories, distribution centers and stores. The company's aim is to reduce carbon emissions and increase its profit margins.
Utilizing data sources from across supply, demand and inventory, combined with Electronic Point of Sale (EPOS) and customer data, the company is using DI to optimize stock levels at each distribution center and coordinate movements of stock between multiple centers, taking into consideration factors such as demand (actual and forecast), production output, processing costs and transportation costs. The solution cut logistics costs by 10% and reduced truck journeys between centers by 200,000km, representing a reduction of 147 Metric Tonnes (MT) in CO2 emissions in the first eight months of its deployment.
Similarly, a leading producer and supplier of aggregates to the construction industry, with a total fleet of 400 vehicles, was able to increase jobs per driver by 15% and reduce mileage by 3% for every job with an automated DI planning solution that predicts job demand and cancellations, maximizes vehicle productivity and plans vehicle routes.
What’s your vision for the future of Peak?
We want to put Decision Intelligence in the hands of every business and build a company people love to be a part of. This means that expansion to support more customers globally is our top priority and we’re expanding in both the US and India, opening Clubhouses in New York, Mumbai and Pune. Sustainable high performance is key to that; we want Peakers to be on our journey for a large part of their careers, we don’t want people who will come in and be burned out within a couple of years. .
We’re investing heavily in R&D following our successful Series C round that closed in August of last year. As we release more exciting platform features and expand around the world, we are excited to see the applications that data science teams outside of Peak develop with the platform – much of what DI is capable of will be discovered in practice.
Thank you for the great interview, readers who wish to learn more should visit Peak.
- NVIDIA: From Chipmaker to Trillion-Dollar AI Powerhouse
- Laura Petrich, PhD Student in Robotics & Machine Learning – Interview Series
- Liquid Neural Networks: Definition, Applications, & Challenges
- Patrick M. Pilarski, Ph.D. Canada CIFAR AI Chair (Amii) – Interview Series
- AI Leaders Warn of ‘Risk of Extinction’