Charles Miglietti Bet That Business Data Needed a Better Story

Charles Miglietti learned the difference between an interesting product and a business before he turned 27. His first company, TL;DR.io, collected human-written summaries of web pages. It was useful, clever and unable to answer the least poetic question in commerce: who pays? After two years, he and his co-founders stopped. Miglietti kept the lesson. A product needs a business model and a customer prepared to exchange money for the result.

That disappointment became the opening move for Toucan. In September 2013, Miglietti began consulting on data visualization. The assignment was not merely to fund an idea in private. It was a market test conducted in public, with invoices. Could a simpler, more designed form of business analytics solve a problem worth buying?

The answer arrived as a $40,000 pilot from a mutual insurance company. There was no finished minimum viable product to demonstrate. The team sold the work, then built the first version in less than two months. It visualized social data for employee representative groups. The first customer was not waiting at the end of the product process. The customer was part of the process.

This is the unvarnished advantage of paid validation. Compliments are pleasantly vague. A contract is annoyingly specific. It imposes a user, a deadline, a budget and a definition of done. For Miglietti, the method offered protection against repeating TL;DR.io: Toucan would discover its company while building its software.

It also changed the quality of the questions. Instead of asking whether executives might like simpler charts, the founders could watch which numbers an actual organization needed, where readers became confused and what had to be explained before a decision could move. Consulting supplied close observation; software made the successful answers repeatable. The arrangement was temporary by design. Miglietti did not want to own an agency that happened to make dashboards. He wanted the customer work to reveal the boundaries of a product. By the end of 2015, Toucan had stopped consulting. Services had served as scaffolding: essential while the structure was going up, and removable once the structure could stand.

Miglietti grew up in Normandy and studied engineering at École Polytechnique from 2007 to 2011, learning programming, artificial intelligence and data analytics. An internship with Apple took him to Portland, Oregon. Entrepreneurship, however, was close to home. His older brother had pointed him toward starting something in technology. Only later did he recognize a longer family echo: both of his grandfathers had been entrepreneurs.

Toucan was formally founded in 2014 with Kilian Bazin, David Nowinsky and Baptiste Jourdan. The company continued consulting while it developed repeatable software, then ended the consulting work in 2015. The founders wanted a product company from the beginning, but they preferred customers to financiers as their first source of fuel.

“We preferred to find customers rather than to go look for financing.”Charles Miglietti, on Toucan’s early years

The path was not a perfectly focused ascent. In its first year, Toucan sold only four times. Early customers came from unrelated sectors. Leads arrived through outbound email, LinkedIn advertising and networking. Opportunity came before pattern. That worked until the company became large enough for inconsistency to become expensive: a salesperson cannot easily repeat a pitch that changes with every prospect.

The team finally examined its own customer base and kept the use cases that repeated. Retail had particular traction; SaaS and human-resources applications also appeared. This was focus discovered empirically, not delivered by a slide deck. Miglietti’s preferred description of his product-management style is one word: “Pragmatic.” It fits because his story repeatedly rewards evidence over elegance.

Pragmatism became most visible when it hurt. At one point, Toucan users asked for a data-entry interface. The request was logical: one cannot tell a data story without data. The team spent months developing it, then concluded that data entry required a different expertise and sat outside Toucan’s essential value. The work was discarded.

There is an important distinction here. The feature was not absurd. It was adjacent. Adjacent ideas are dangerous precisely because they arrive carrying reasonable arguments. Miglietti’s response was to buy existing solutions when a supporting capability already existed and reserve Toucan’s effort for the experience it could make its own.

That experience sharpened a belief that product teams should not repeatedly construct undifferentiated dashboard systems inside their own applications. Toucan could provide the user-facing analytics layer while its customers kept engineers on the problems unique to them. The company’s product was also an argument about opportunity cost.

Even the name made the argument visually. Miglietti and his co-founders wanted a mascot that could break with the cold conventions of traditional business intelligence. A bird suggested agility, a high vantage point and movement. The toucan added color and a pleasantly exotic association. Business software often dresses like a tax office to prove that it is serious. Toucan arrived in feathers.

The whimsy masked a precise thesis: analytics for non-specialists must be inviting enough to use and clear enough to act upon. Data visualization was not the ornamental layer after analysis. It was the means by which analysis crossed the final distance to a decision.

By 2019, Toucan reported $5 million in annual recurring revenue, roughly 70 employees and a product the team believed was ready to scale. Only then did it raise a large institutional round, described in contemporary accounts as $10 million or €8 million depending on currency and announcement. The money was earmarked for international hiring, including the United States and the Netherlands, and for deeper customer work.

Miglietti moved to Boston in early 2020 to support the American expansion. Lockdown followed almost immediately. His account of the period avoids heroic varnish: he worked from a home office, spent time with his wife and three children, explored the outdoors around Boston and took guitar lessons. The ordinary details give the move its correct scale. Companies cross borders through strategy; people still have to make a home.

His route from technical co-founder to chief executive passed through product ownership. He began driving Toucan’s roadmap because the work needed an owner, eventually serving as both CEO and chief product officer. Asked about his interests away from work, he has named cooking, music and hiking. These are tactile occupations. Ingredients, chords and paths give feedback without scheduling a quarterly review.

A decade after Toucan began, the interface around data shifted. Customers increasingly asked for an AI roadmap. For nine to twelve months, the answer remained “coming soon” while nearer-term priorities won. Meanwhile, AI-native products made a conversational experience feel less like an experiment and more like a new expectation.

Miglietti and the team reached an uncomfortable architectural conclusion. Adding a durable AI experience to the legacy product would take quarters. They chose to rebuild the product from scratch around conversational analytics, while continuing to maintain the earlier platform and construct a bridge for customers whose needs were different.

The new entry point reverses the old sequence. Traditional analytics begins with charts, dashboards and filters. Toucan’s newer bet begins with a question. The system interprets it, applies governed business definitions and returns a visualization in the conversation. In 2026, Toucan also introduced an MCP connection designed to let an existing AI assistant query data and return a chart while respecting tenant boundaries and the company’s semantic layer.

That final qualification matters. A conversational interface can feel effortless while the work behind it becomes more exacting. Customer-facing analytics has to know whose data it may retrieve, what a metric means and how to show the result without inventing confidence. A wrong answer in a demo is embarrassing. A wrong number placed inside a customer’s product is operational.

The interface changed. The old question survived: can a non-specialist understand what matters?The continuity inside Toucan’s pivot

Miglietti has recently made a parallel argument about product management itself. AI may help engineers ship faster, but faster production increases the pressure to accept every plausible request. The feature-factory risk rises when development becomes cheaper. Product judgment, in that world, is not automated away. It becomes the scarce part.

This returns neatly to the data-entry feature Toucan threw away and the first startup it shut down. Tools become faster. The founder’s work remains selection: which problem deserves a company, which customer reveals the pattern, which capability belongs inside the product, and which architecture has become a limit.

Miglietti once described himself as passionate but not impetuous, willing to take risks without skipping steps. Toucan’s history supports the distinction. The company validated through service work, productized what customers bought, raised major capital after reaching meaningful recurring revenue, crossed the Atlantic when its sales motion required it and rebuilt when the old foundation could no longer carry the new interaction.

The toucan still supplies the color. The operating philosophy is plainer: make complexity legible, and let evidence edit the plan.