CrowdFlower’s Chief Executive Officer, Robin Bordoli gives AI Business’ readers an insight to how they have implemented AI in their industry, their uses cases for AI solutions and how he sees the technology developing in the future. CrowdFlower is the essential human-in-the-loop platform for data science teams. CrowdFlower helps customers generate high quality customised training data for their machine learning initiatives, or automate a business process with easy-to-deploy models and integrated human-in-the-loop workflows. Their platform supports a wide range of use cases: self-driving cars, intelligent personal assistants, medical image labeling, content categorisation, product categorisation, and search relevance.
Robin Bordoli is Chief Executive Officer at CrowdFlower having joined the company in February 2015. Robin has spent the past two decades helping high growth companies launch and scale platforms and products into rapidly transforming markets.Prior to CrowdFlower, Robin was the Vice President & General Manager, Strategic Consumer Industries at Marketo where he launched and led the business unit selling to consumer digital marketing teams. Robin has also held leadership roles at Jive Software, Worksimple, Yahoo, Excite@Home & Micromuse.Robin holds a Master’s degree in Engineering from Cambridge University and a Master’s degree in Business Administration from Stanford University. Outside of work Robin spends his time trying to keep up with his two young children and enjoying all that the Bay Area has to offer.
AI Business asked Bordoli to outline CrowdFlower’s key proposition in enabling an AI-powered business, to get an insight into why they have chosen to apply this technology?
Bordoli explains that in general, this allows CrowdFlower to reduce the time to deployment for and increasing the commercial viability of machine learning for an enterprise. “Our platform, in particular, does this in two ways”, Bordoli explains, saying: “First, we generate customised high-quality human-labeled training data from unstructured text, images, audio and video from which initial machine learning models can be created”.
“A machine learning model without training data is like a car without gas. It’s just an expensive object that won’t go anywhere”
“Second, we incorporate human-in-the-loop workflows to handle the low confidence outputs from a machine learning model that needs human review”, Bordoli explains.” A machine learning model without human-in-the-loop will either remain a science experiment because the accuracy is too low for real world applications or lead to bad outcomes because too much trust is placed in the model”.
Mentioning the human-in-the-loop platform, AI Business asked if Bordoli could outline where and how their products can be applied?
“Our human-in-the-loop platform is used in a wide variety of use cases and verticals”, Bordoli explains. “Some of the most popular ones are dense image labelling for self-driving cars, medical image labeling for healthcare companies, conversation categorisation for intelligent assistants, content categorisation for media companies, customer support ticket classification for technology companies, social data insight for consumer companies, CRM data enrichment for b2b companies, product categorisation and search relevance for e-commerce companies”.
In other words, the platform has a large range of use areas and can be applied in various industries.
Having seen the benefits of applying AI in their business, we asked what CrowdFlower’s enterprise strategy is, in the short-and long-term? Are there any particular industry verticals you are focusing on for your solutions going forward, or any particular applications on the horizon?
“Our enterprise strategy is to provide an enterprise-grade SaaS platform that is used by data science teams at Fortune 2000 and fast-growing data-driven organisations to generate the high quality customised data they need”, Bordoli outlines.
“We’ve executed on this strategy across a variety of industry verticals such as automotive, media, e-commerce, technology, and financial services. Our long-term strategy is to be an essential part of the AI ecosystem by partnering with the major machine learning platforms”, he says.
AIBusiness wanted to know whether Bordoli has experienced any challenges when looking to implement their products and solutions in the enterprise?
“One of the biggest challenges for us when helping data science teams understand and adopt our platform is making sure their mindset is one of iteration rather than linearity”, Bordoli explains. “Some data teams have a mindset of “once and done”, but in reality when data scientists address a new problem in a new domain there are many “unknown unknowns”, he says.
So what are Bordoli’s recommendations to tackle these issues? “Rapid iteration on our platform through collecting human-labeled data in minutes and hours rather than weeks and months is both possible and highly advisable to improve scope, and ultimately the quality of the resulting data”, he says.
“It’s our experience from working with some of the biggest and best data science teams that an iterative rather linear approach yields the best results”.
Looking ahead towards future project, what new products and solutions can we expect from CrowdFlower in the immediate future?
“We recently launched a new product called CrowdFlower AI powered by Microsoft Azure Machine Learning”. Bordoli explains that they partnered with Microsoft to bring this product to the market to serve companies who wanted the benefit of machine learning to automate business processes, such as customer support ticket classification who didn’t have the inhouse machine learning expertise to deploy themselves.
“You should expect to see further innovation and partnerships that democratize machine learning and make AI possible for every business and not just the technology elite”.