Tom Evslin had an oddly specific lens for a company building software for “the internet thing” in the early 1990s: he was an email guy. At Microsoft, while the company worked on Exchange Server and Outlook, Evslin described himself and others as “internet radicals.” The friction was not about whether email mattered. It was about when the internet should matter inside Microsoft’s product roadmap.

In Evslin’s telling, Bill Gates was willing to pour effort and money into Exchange because Lotus Notes could become a platform the way Windows was. Gates was afraid Lotus Notes would threaten Microsoft. Evslin agreed Exchange needed to “fend off the threat,” but he also says the internet radicals had a different priority. They wanted upcoming products delayed until they were relevant to the internet. Gates disagreed. Evslin describes Gates as saying Microsoft was likely already behind schedule, because the company was “always looking for an excuse to be late,” so Microsoft should launch the next products, including Office, Exchange, and NT, and “worry about the internet afterward.” Evslin calls that a mistake, and the timeline he lived through suggests why it’s the kind of strategic dispute that can change outcomes fast.

To understand the stakes, you have to see the corporate incentives in play. Lotus Notes was competing as an enterprise platform, not just as a messaging tool. Exchange, by contrast, was designed as Microsoft’s answer for email and collaboration, but email is only valuable when it can move across systems. Evslin was responsible for gateways in Exchange, the pieces that connect one mail system to another. Some gateways connected to services like CompuServe. Others linked to MCI Mail, which Evslin says he had worked on as a client. And then there was the SMTP gateway. In the early days, SMTP mail used on the internet “wasn’t widely used,” at least by the audience Microsoft expected to serve.

Then the demand curve shifted anyway. While Microsoft’s corporate clients were insisting that internet communication was “not safe” and “not secure,” Evslin says the internal posture got overwritten by customer behavior: “All of a sudden we got a huge number of requests for this SMTP gateway.” When Evslin looked into it, he found more and more email was going over the internet. The official stance might have been “nope,” but engineers were doing the communicating work. Evslin says their engineers were communicating on the internet and either didn’t understand it, or didn’t at the time. This is the part that should make modern executives uncomfortable: not because Microsoft lacked foresight, but because the organization’s external messaging and internal reality were drifting apart.

Once you have gateways, you create a bridge to an ecosystem. And ecosystems do not wait for internal debates. That pressure also shows up in how Outlook was being shaped. Evslin says the early client for Exchange was built to support graphics and fonts, so you could format an email similarly to a Word document. Where things “didn't go far enough,” in his view, was that Microsoft still had two separate formatting engines, so the Word and Outlook client never merged the way he thought they should have. That mismatch matters because user experience in enterprise software is often a hidden adoption lever. If formatting, collaboration, and familiarity are off, even the best backend strategy can stall.

Evslin’s story also moves sideways into AT&T, which adds another layer of strategic lesson. Soon after Gates chose to launch products before deep internet integration, AT&T invited Evslin “to come and develop their internet strategy.” At that time, Evslin says AT&T didn’t have a clear plan for the internet. The company had tinkered with proprietary networks, but Evslin reckoned AT&T should become an ISP. He also pushed for all-you-can-eat pricing, which smaller providers had tried but AT&T had not yet launched. Internal debate included the argument that “you can't launch a new service in less than seven years.” Evslin’s counter was blunt: the solution to that kind of timeline is to launch it fast and then adapt. WorldNet, his AT&T project, offered straightforward internet access to customers familiar with portals and proprietary networks, with the promise to avoid the “heart-stopping” telephone bills.

WorldNet’s demand grew so quickly that AT&T had to control sign-ups to keep the service from earning the “America On Hold” nickname sometimes applied to America Online. It’s a classic scaling problem, but it’s also a signal: when customers are ready, the bottleneck becomes operations, not ideas. And when that bottleneck hits, it changes what leadership measures. It shifts the focus from “does the strategy make sense” to “can the system handle what the market is already doing.”

With the internet in the ascendant, Evslin became interested in another technology that connected directly to enterprise and consumer expectations: voice over IP. He acknowledges it was an “impossible sale” inside AT&T. So in 1997, Evslin moved on again, founding ITXC, a wholesale VoIP carrier. The through-line is timing and adoption friction. Internet-driven demand surfaced fast, customer behavior overrode internal caution, and organizations that tried to sequence “internet later” into their product and pricing strategy got dragged toward integration by the real world.

For today’s executives, the second-order implication is not just historical trivia. It’s a governance question: how do you prevent internal risk narratives from lagging the external behaviors your customers are already taking? Evslin’s account suggests that even when your official stance says “not safe” or “we'll integrate later,” engineering work and customer requests can force your roadmap to follow the internet anyway. The winners are often the teams that notice that shift early, build the bridges (gateways and standards support), and then adapt faster than the enterprise can say “we're not ready.”