Declara launched in 2012 to considerable funding and fanfare—including a shoutout from former President Barack Obama. But its AI-powered social learning platform struggled to live up to expectations, and never quite found its footing in the education market.

Instead, it has caught the attention of a European buyer: An Italian blockchain company that builds software to support the “Internet of Things” ecosystem has acquired Declara. Financial terms of the deal were not disclosed.

It’s an unusual buyer for an education technology startup. According to Nelson Gonzalez, Declara’s co-founder, the buyer was drawn primarily to Declara’s machine learning technologies, which it hopes to roll into its blockchain software.

But Declara is not going away. Gonzalez assures that Declara’s educational offerings and clients will be supported. “We’ll continue to serve educators all over the world, and foster better enterprise learning and development.”

Based in Palo Alto, Calif., Declara’s rapid rise to fame fed into some familiar Silicon Valley startup tropes. There was a harrowing comeback story, featuring its co-founder and first CEO Ramona Pierson, who suffered a near fatal accident caused by a drunk driver. As she recovered, she continued her work in applying neuroscience research and artificial intelligence technology in education. She started and sold one edtech company, SynpaticMash. Declara was the second.

There was also a litany of buzzwords in the company’s pitch, as Declara claimed to mash up machine learning, artificial intelligence, semantic analysis and neural networks together into an “intelligent social learning platform.”

Declara had two core components: a “cognitive graph” that could analyze behaviors and patterns in how users on its platform accessed and shared content, and interacted with one another. This data is then used to create learner profiles for each user.

The second was Declara’s ability to ingest, index and tag digital materials that its customers wanted on their platform. (It could also do this for openly-licensed content publicly available online.) Based on users’ learning profiles, Declara claimed it can predict and automatically recommend the right content tailored for each user.

Among Declara’s earliest clients were education departments and ministries that used it for teachers’ professional development. Its first customer was Australia, which contracted with Declara to build a platform to help its teachers implement a new national curriculum. Similar large-scale deployments followed to support teachers’ professional learning in Brazil, Chile, Mexico and Puerto Rico. For California’s department of education, Declara built a social network to help teachers share professional learning resources.

Declara also white-labeled and sold its platform to companies, including 3M, Genentech and Renault, which used the platform for employee training. The majority of these deals were pre-paid, multi-year contracts involving seven and eight figures, according to Gonzalez.

Big Capital, Bigger Expectations

Those early successes attracted an influx of capital. Declara raised $5 million in seed funding in 2013, quickly followed by a Series A that stretched multiple closings. In total, the company has raised $38 million.

But the rapid influx of money also came with “massive expectations” from investors for scaling the business that the company struggled to live up to, according to Gonzalez. Despite the company’s early success inking deals with big clients, they did not translate to repeatable sales. Those deals were largely driven by the relationship that he and Pierson had. “We expected our salespeople to do some magic to replicate the early deals. But that’s not realistic…[since] they were largely founder-driven,” he adds.

The company struggled to understand who its target customers should be, and what pain points Declara was trying to help solve. Across its education and corporate clients, the use cases were different enough that it was difficult for the company to identify its target market or develop customer personas. As a result, Declara was “a solution in search of a problem,” says Gonzalez.

“The cognitive graph [that the company built] is a powerful tool to understanding human behavior,” he says. “But for what? By itself, it’s a solution to an unspecified problem, and that makes it tough to scale.”

As a result, the team found itself proactively seeking customers to find use cases for the Declara product—not the other way around. “The definition of product-market fit is when the product is selling itself. Post Series A, the product needs to sell itself. You shouldn’t need the founders to do it.”

Complicating matters was that Ramona Pierson, its iconic co-founder and CEO, fell ill with a medical complication that forced her to leave Declara in 2017. She later took up a post at Amazon.

Declara had worked with 20 customers over its history, and a handful remain active clients. But it has not notched a deal in some time. Its most recent customer was the Suncoast Campaign for Grade Level Reading, which contracted with the company in 2017 to build a social learning platform.

Declara tinkered with plans to develop a slimmed down version of the platform to sell directly to consumers. It also thought about pivoting to focus exclusively on the corporate learning market. Closing shop was another option on the table.

What did scale was the company’s headcount, which at its largest ballooned past 120 full-time employees with satellite offices in Mexico and Singapore. The company has since trimmed down its staff considerably.

Despite its struggles, Declara was fielding inbound interest from acquirers starting in 2016. Those conversations ultimately led to its eventual acquisition.

Gonzalez, who will remain as an advisor to Declara, says he’s been humbled by the journey as a hot-shot, overcapitalized startup that never found its footing. “Forty million dollars is a lot of money to have before you have product-market fit, and when you have a mandate from investors to scale.” He’s now imparting these lessons learned through a course he’s leading at the University of Pennsylvania’s Graduate School of Education.