Twill founder and CEO Michael Wright is using his payments background and technological breakthroughs to bring the benefits of AI [artificial intelligence] and data analytics, previously only available to large companies and small- and medium-sized businesses. By consolidating data and providing enhanced analytics, Twill will provide a more efficient experience for business owners who use an average of 16 SaaS products to manage daily operations.
When Wright co-founded Payment Cloud nine years ago, he admittedly knew little about payments. That quickly changed as he built the sales team and 1099 agent channels before designing products for that channel.
“With payments, you get a really good view into the entire business you’re talking to,” Wright said. “I’ve had thousands of conversations with merchants.”
Siloed data was one clear issue. Those 16 systems the average owner was using didn’t talk to each other, and owners had no time to fill those considerable gaps.
“That’s a solvable problem,” Wright said. “I think we can address that.”
How new technologies make Twill possible
Wright describes Twill’s evolution as the convergence of multiple technological waves. Universal APIs have grown in popularity, allowing developers to design a single API.
Artificial intelligence provides the means to allow non-technical users to thrive within a technical product.
Large companies can deploy data science teams to design systems and leverage the data they generate. The mom-and-pop can’t. Enter Twill.
“What we do at Twill is combine those things,” Wright explained. “We allow SMBs to connect QuickBooks, Shopify, their POS system, marketing spend…We bring that data in, and we have a conversational, generative interface that allows them to interact with their data.”
How Twill addressed tech challenges
While it’s not technically difficult to integrate with an API, repeating the process with hundreds of them takes much work. Twill connected with Rutter, a provider of unified APIs. When Twill users log into, say, QuickBooks, the API pulls the necessary information.
That’s the front end. Designing back-end architecture was trickier. Definitions aren’t universal. Banks and accountants do things differently. Twill had to make the right connections between all the data coming in.
“That was important to us because we’re not connecting with people who know how to architect a database,” Wright said. “So we have to do it for them.
“And we’re working with an AI that needs to know a lot about the database. It has to know what these terms mean and how they relate to one another so we can create these insights.”
Having realistic AI expectations
Wright described the Twill team as tech realists. They don’t expect AI to do everything.
That’s important because whenever the latest tech fad arrives, companies dive in. They seek to capitalize on its novelty, which often captures VC attention. If they get lucky and the tech is truly transformational, maybe they will become a giant in the field.
Wright has tempered expectations. He tested ChatGPT and Anthropic, asking them business questions and uploading spreadsheets.
Wright received meaningful responses. With the right prompts, AI systems can parse data and provide useful answers. Twill’s task was to develop a layer that simplifies complex tasks for users.
People mustn’t see AI as a cure-all, especially if they’re designing a system like Twill’s. Wright said AI isn’t that effective at doing things yet. It may not give the same answer every time because it takes different routes to answers.
But Twill needs the same answer every time, so it built a system around agents instead of letting an AI system do everything. Play to AI’s strengths of understanding what is being asked.
“We don’t even want to go to market as an AI company, to be honest,” Wright admitted. “Ai is in the background. We’re building a platform that is going to work for you. Yes, we use a little bit of AI to help, but we’re not going to be in your face about it.”
With AI being so new and many improvements on the way, how does Twill plan for new capabilities that come 12, 18, or 36 months from now?
Wright said it’s fortunate both Twill and AI are in their infancy. As AI improves, so will Twill. The design team has a long list of enhancements awaiting that growth.
Twill’s UX
Great care was also taken when designing Twill’s user experience. It had to be easy and intuitive. After signing up and logging in, users view Twill’s connection library. They click on the systems they want to connect to and log in.
Once the data is loaded, Twill will give prompts of databases to add and questions to ask. They are designed to improve the user experience. A chatbox can provide line-by-line insights.
As dashboards develop, an SMB owner can look at a dip in a sales chart and find answers to why.
“That’s the real power,” Wright said. “We drill on the specific reasons for why this is happening so you can take specific actions.”
With so many integrations, Wright said security is a significant concern. Crucial is building customer trust so they’ll connect systems and make them useful.
Many people fear AI systems will take their data and learn from them. Wright said in Twill’s system, AI never accesses business data. Agents use it, not AI. Twill is SOC2-compliant and has all relevant certifications.
Wright said payments weave all business areas together and drive business. There are opportunities to derive valuable insights from deep layers of business data. Using an AI layer, a system can analyze invoices and schedule payments.
As those systems grow, Wright said they must adapt to evolving needs.
“The industry is changing,” Wright said. “No one wants just one kind of payment anymore; they want to do everything, and they don’t want 17 different products to do everything.
What’s next
Twill raised a pre-seed round in April 2024. Wright said the company is on track to release an MVP by Halloween. Twill is also working with a small group of users and hopes to expand a beta to 50 owners in 1Q25 before opening up more in mid-2025.
