AI Tools for Savvy Startups

Machine learning, natural language generation and related technologies will transform how new ventures get off the ground
By: 
Alan Morantz
Toolbox with tools on laptop keyboard.

Garry Kasparov never had a chance. In 1997, the world chess champion lost to Deep Blue, a computer powered by artificial intelligence (AI). In the match, Kasparov was evaluating three chess moves per second. Pretty impressive. In that same time, Deep Blue was plotting more than 200 million moves.

Imagine you’re an entrepreneur today. What sort of edge would you gain by having Deep Blue’s smarter cousin as a consultant on your senior management team? How many more viable business ideas would you generate? How much faster would you be able to turn those ideas into money-making products or services? Would your decisions finally be based on cold, hard data rather than biased and blinkered thinking?

This isn’t a fanciful scenario. AI-powered applications are tailor-made for startups, and many are already within reach. The magic in these technologies—machine learning most notably—is less about intelligence and more about pattern-seeking and prediction, the building blocks of innovation. That’s why some people refer to AI as an “invention of a method of invention”—one that is particularly attuned to innovating entrepreneurs.

AI can improve each step of the entrepreneurial process, from discovery of an opportunity to its development and exploitation. The requirements are fairly straightforward. As Steve Thomas, executive director of the Analytics and AI Ecosystem at Smith School of Business, points out, small businesses can reap the benefits of AI as long as they have deep and clean data sets of customer interactions and access to cloud-based platforms that offer AI applications, such as Amazon Web Services, Microsoft Azure and Google Cloud AI. These platforms offer flexibility and scalability; entrepreneurs can experiment with machine learning services free for a period and then opt for a pay-as-you-go plan for specific projects such as building virtual assistants or running a sentiment or image analysis project (usually charged per hour of computing time).

Here’s a closer look at how AI technologies can power various stages of the entrepreneurial journey.

Building a startup team 

Nascent entrepreneurial enterprises are often thin in human resources expertise. At some point, though, they have to recruit a team of functional experts—people who can’t be found among classmates or within a family social network. AI is particularly promising in this area. 

AI firms can help clients to eliminate biased language in their job descriptions, screen resumés using natural language processing and machine learning and manage the candidate experience using chatbots. One firm, Pymetrics, puts a firm’s employees through a series of tests resembling video games. Then it feeds the data into an algorithm to build a persona of the people who fit a particular role. Candidates then take similar tests to see if they’re a match in behaviour and team culture.

Prospecting for ideas

AI-powered tools can connect disparate pieces of information to identify market needs or failures that can be parlayed into novel entrepreneurial opportunities.

Social sentiment analysis and natural language processing, for example, can be deployed to analyze social media and online customer forums for a product or service category that is ready for disruption. A sentiment analysis tool is software that analyzes text conversations and evaluates the tone, intent and emotion behind each message. Essentially, these faint signals can be fed into an algorithm to identify counterintuitive insights.

If entrepreneurs are stumped for a startup idea, they can visit IdeasAI. This site uses deep learning to generate new product and business ideas. Visitors “like” or “dislike” an idea, and in the process train the algorithm and improve future ideas. 

Designing and prototyping  

AI can disrupt and accelerate the product design process by offering a more rigorous and bias-free assessment of data and by integrating multiple sources of information. A modest example is IntelligentX, a British startup that used AI to develop a new line of beers. The firm first deployed a bot to ask consumers about flavour preferences. It then fed the data into an algorithm that was used to create its first line of beer products. The recipes are continually tweaked based on ongoing feedback.

Firms such as Autodesk are pioneering AI-powered tools for generative design. Designers or engineers input design goals into the generative design machine learning software, along with parameters such as performance requirements, materials and cost constraints. The software explores potential solutions, quickly generates design alternatives and learns from each iteration what works and what doesn’t. The software also generates digital models for rapid prototyping via 3-D printing.

Selling products and services 

AI-enabled sales tools can free up salespeople for higher-value customer-facing tasks by automating lead engagement and qualification and by notifying salespeople about who is worth contacting. Or they can superpower salespeople with AI systems based on social signal processing and computer vision that can interpret behavioural cues, helping salespeople to evaluate customer reactions to product features or prices. Or they can replace salespeople with chatbots created from natural language processing and deep neural networks

AI can also unlock marketing information that may be hiding in plain sight. IBM's Watson Personality Insights platform scans a startup’s social channels to create personality profiles of target customers. It can then identify influencers or members of customers’ peer groups who could become company champions.

Scaling the enterprise

When administrative, design and sales processes are largely driven by AI applications, it becomes much easier to scale an operation once the new business catches on. Dealing with resource constraints becomes less about adding staff and more about building the computing capacity to deliver more AI functionality.  

Automating back office functions, customer relationship management and other routine tasks means more human time on higher value-added work. 

If chatbots prove effective, they can easily be added to. Top salespeople can even be cloned. There are firms that use conversational AI techniques to analyze top-performing salespeople in a client’s organization. They use that data to train machine learning systems and create virtual selling assistants that can respond to buyers’ questions 24/7. 

Developing future entrepreneurs

Researchers are using AI to better understand the dynamics of entrepreneurship. In the years ahead, we can expect to read about studies that advance our understanding of startup and investor decisions in board meetings, product development sessions, entrepreneurial pitches and other settings. These studies will be based on social signal processing that analyzes body postures, facial expressions and vocal behaviour cues in video clips and images.

One study already published used machine learning to predict the outcome of crowdfunding campaigns. Using text, speech and video-related meta-data in 20,188 crowdfunding campaigns, the researchers built a model that was 73 per cent accurate in explaining campaign success or failure. 

Another study used an AI-based technology to analyze the “digital footprints” of entrepreneurs to gain insights on the personality characteristics of superstar entrepreneurs. The researchers compared the personality characteristics of 106 of the most influential business leaders by analyzing their Twitter messages with a computerized text analysis tool. The study showed that superstar managers are more entrepreneurial in many personality characteristics than superstar entrepreneurs. 

None of this is to suggest that using AI will be easy for entrepreneurs or their teams. It won’t be. AI technologies demand a certain level of data management maturity. Cloud-based AI applications present their own challenges, chief among them privacy, data ownership and data portability. 

Notwithstanding these issues, it is certainly tempting to imagine the age of AI as the most interesting time to be an entrepreneur.

Smith School of Business

Goodes Hall, Queen's University
Kingston, Ontario
Canada K7L 3N6

Follow us on:

Queen's logo