We live in the age of artificial intelligence. As tools such as the ChatGPT chatbot startle us with their “human-like” capabilities, corporations are quietly but rapidly embedding artificial intelligence and machine learning technologies across their mission-critical customer interactions.
Missing from the revolution: Ways to bring the power of machine learning down to the desktops of most employees. But Akkio, a STEX25 firm, is now delivering artificial intelligence and machine learning software for anyone in an organization who routinely analyzes operational data.
“Akkio is built for anyone in a business who's working with a data-driven workflow, trying to extract value from the data,” says Jon Reilly, co-founder and chief operating officer. “If you understand working in a spreadsheet, you can now leverage artificial intelligence.”
“Whatever your business is built on, whatever your key performance indicators are, we make it really easy to use machine learning models to understand what's driving them, and then optimize them so that you can do better as a business at executing,” Reilly says. “Akkio makes it easy to do bottoms-up machine learning, to get your entire organization transitioned into using machine learning on a daily basis. We're all about getting quick wins with machine learning.”
Working with workflows
Akkio's core targets are the employees who work with spreadsheets and tabular data to understand business operation problems in sales, marketing support and finance, Reilly says. Traditionally those efforts involve a lot of manual manipulation—entering complex formulas, filtering the results to get the view you want, graphing your findings and trying to figure out what's happening.
With Akkio software, you start by uploading a file of tabular data such as an Excel worksheet or directly connecting a live data source such as Salesforce or Snowflake. Next you prepare the data for analysis. The software accelerates this normally tedious process with a natural-language-based data preparation feature that lets you quickly do complex calculations and apply filters to pull out what interests you.
Once your data is prepared, “in the click of a button you can train a machine learning model to predict your key outcome,” Reilly says. For instance, you can look at sales leads to find out which leads panned out. “We'll learn from every other feature you have in your data about those leads, to figure out what the patterns are,” he says. Next, the software presents easy-to-understand graphics that show which features are most likely to encourage a given sales prospect to buy the product.
In the click of a button you can train a machine learning model to predict your key outcome
Once these patterns emerge, it's simple to deploy the machine learning models to automate decision-making in workflows, he says. For instance, you can take a new sales lead, run it against the model and quickly see the probability that the lead will turn into a sale and, if so, the value of that sale.
Additionally, Akkio software can take on larger jobs such as forecasting revenue, picking out the driving factors, examining how those factors are trending and seeing how they will impact company revenue.
Crucially, the package provides estimates of the reliability of its analyses. “Machine learning is about probabilities,” Reilly notes. “By playing the more likely outcomes, you won't always win, but you'll win enough to add up to a winning season for your team.”
Insights across industries
The software is designed to provide insights from operations at any scale. “If your pattern is simple and straightforward, and you're just trying to automate a repetitive task, you can get by with as little as 1000 examples of that task to teach the machine learning engine,” says Reilly. At the high end, other customers create models on hundreds of millions of records.
Many companies use Akkio to predict financial outcomes, such as trends in revenue, including aspects such as seasonality and return on investment for various expenditures.
Other successful applications vary tremendously, Reilly emphasizes.
One giant firm examines current interactions on its website, predicts the lifetime value of each interaction and uses that information to optimize its advertising dollars in real time, he says. An international shipping company employs the software to analyze operations so its customers can pick the best shipping routes, minimize costs and understand the probability that a package will arrive on time. A consumer electronics firm slashed the time needed to review feedback on its new product development by 75%. One firm that raises funds for political candidates applied the software to ranking potential donors and more than doubled the contributions raised. A medical device manufacturer found that it could predict which clinical trials would fail with 90% precision.
Akkio software also enables near-instant response to online events. For instance, “you can build sentiment analysis, where you take a tweet and evaluate how likely it is to be positive or negative,” Reilly says. “Then you can monitor mentions of your business on Twitter. When someone says something negative about your business, you can have somebody interact with them and try to fix the problem. Or you can monitor your competitor's Twitter, and when someone says something negative about their business, you can reach out and get a new customer.”
You can build sentiment analysis, where you take a tweet and evaluate how likely it is to be positive or negative
Bringing machine learning home
Reilly, chief executive officer Abe Parangi and G&A head Craig Wisneski previously worked together at a 3D printing company in Boston, where they struggled with common business challenges such as optimizing the flow of sales leads and responding to customer support requests.
“Machine learning is a perfect pairing to solve these types of problems,” Reilly says. “But when we surveyed the market, we couldn't find any tools that would enable us to do that with our own data. And that's a problem, because your data is something that only you know really well.”
The Akkio founders realized that a machine learning platform that made it easy for non-technical people to examine their data and leverage it in immediate decision-making could offer massive improvements in efficiency. “That was the seed of the idea,” Reilly says.
Founded in 2019, Akkio employs 20 and is based in Cambridge. Wisneski earned undergraduate and master's degrees at the MIT Media Lab, and many Akkio software engineers are MIT alums. The startup gathered a $3 million seed investment in 2021 and completed its Series A round of financing in January.
Akkio commercialized its software in the last year but has been testing the platform since it was built. Thousands of people have checked out the software, and “they break it in ways we didn't even imagine when we set out to design it,” Reilly remarks. “We fix every way that they break it, we learn to handle different data structures and formats and messy data, and it makes the product more robust.”
Over the next year, Akkio will extend its platform with dashboards that let you immediately see how all of your key performance indicators are trending and the factors that are driving the trends, he says.
Akkio competes in machine learning tools for the masses with numerous other startups and with giants such as Google, Amazon, and Microsoft. “The axis on which we differentiate and win is our ease of use and simplicity of adoption,” says Reilly.
“We're the tool that you'll use to optimize your marketing team, your customer support team, your sales operations team, the long tail of your operational business that can also be optimized with machine learning,” he says. “Every business will need to adopt machine learning over the next five years in every workflow if they want to remain competitive. Akkio makes that easy to do.”