The accidental inventor: how Richard Tworek went from fixing submarine software to building Riverbed’s autonomous enterprise
Richard Tworek, Chief Technology Officer of Riverbed Technology, has spent more than 30 years treating technology as one continuous puzzle, and when he is asked what connects a career spanning three chief executive roles, several startups built and sold, and a period commanding software aboard a nuclear submarine, his answer does not move, because for him the work has only ever been about one thing.
“It is the same thread all the way through,” he said. “You see a problem and you use technology to solve it.”
The instinct surfaced before he could properly speak, and he tells it with the ease of a man who has told it many times, recalling the photographs his mother kept of him as a toddler. “I have pictures of me holding the phone, trying to dial it,” he said, describing a boyhood spent dismantling radios to understand how each part made the others work, a curiosity he has never quite been able to switch off.
He is candid that none of it was plotted in advance, and that the family expectation pointed in an entirely different direction. “If I had listened to my mother, I would have been a doctor,” he said. What pulled him elsewhere was not ambition so much as temperament, a disposition he can only describe in general terms. “It is a natural curiosity about how things work,” he explained, and the reward, as he describes it, has always been the moment a solution assembles itself in his head. “You have that moment where you realise that if you take this piece and this piece and put them together, you can do something really useful.”
The first lesson came from a lawn, not a laboratory
By the age of 12 Tworek was already running two businesses in parallel, repairing his neighbours’ television sets and cutting their lawns, and it was the grass rather than the electronics that delivered his first lesson in capital. He noticed that the model homes used by local real estate agents were being neglected, and he pitched the agents directly on fixing the problem. “I told them I could make it look nice,” he recalled, and the arrangement grew until it was close to a full-time occupation for a teenager.
What he did with the proceeds is a lesson he still returns to with some rue, because he spent rather than reinvested. “Instead of spending money on hamburgers and whatever I wanted, I should have put it into the business and bought a trailer and more equipment,” he said, before delivering the punchline he has clearly polished over the years. “I would have been a lawn billionaire by now.”
University was supposed to make an engineer of him, and instead it quietly redirected him toward the discipline underneath engineering, as he kept enrolling in mathematics courses without fully understanding why. The reason eventually became obvious. “After four years of mathematics, I worked out that what I actually liked about it was the ability to solve problems,” he said. Aware that a pure mathematics degree offered few exits, he took a master’s in nuclear engineering at the Naval Nuclear Power School, a qualification that placed him aboard a United States submarine as an officer and, almost incidentally, produced his first piece of shipped software.
Richard Tworek, CTO, Riverbed
An idle computer produced his first product by accident
Serving as a department head in the submarine force around 1980, Tworek noticed both an operational problem and an unused computer sitting close at hand, and the combination was irresistible to him. “I thought I could solve the problem, and I did,” he said. The software he wrote went on to run across the submarine fleet in the Atlantic, although he still declines, decades later, to describe what it actually did. “It was my accidental first product,” he said.
The deliberate version arrived when a contract to build document systems for government institutes left him unable to find any existing tool that did what he required, and his reaction was characteristic. “I looked at it and thought, this cannot be that hard,” he said, so he built the system himself, won a run of contracts, and found curious visitors filing through his office to see the work until the commercial implication announced itself. “That was when the light came on, and I realised I could sell this,” he said.
He ran the resulting company out of his own basement and grew it to roughly 130 people across consulting and software before selling it in the mid-1990s, and he built it on a hybrid model he still describes with evident satisfaction, shipping most of the product and charging customers to finish it. “I would build 80% of the software, and for the remaining 20% the customer would engage my professional services to tailor it to their environment,” he said. “It was phenomenally profitable.”
A run of rescues and reinventions, each turning the same screw
The decision to sell was less a strategy than a symptom of restlessness, which is a pattern across his record. “I simply grew tired of it,” he said, before becoming a major shareholder and board member of the public company Infodata.
Roles at Merant, at Nortel, and at a venture-backed voice-over-IP company eventually sold to Cisco followed, and the brief never really changed, because in each case he was brought in to find the common thread between disparate technologies, correct the market the business was chasing, and adjust the technology to fit. At Nortel, handed an open mandate by the chief executive of what was then a $10 billion company, he applied the only method he trusts. “I was told to go and figure it out, and so I did what I always do, which is look around and work out where the problem is,” he said.
The problem he found was that enterprises were accumulating incompatible phone systems through endless mergers and acquisitions, and his answer was to invent the click-to-call capability that is now taken entirely for granted, software he believes remains in use. “Nortel eventually went bankrupt,” he said, “but the technology outlived the company.”
The book business taught him where value actually sits
The most revealing chapter in his record has nothing to do with networks at all, because at one point Tworek left technology altogether to build a book business alongside his wife, acting on the conviction that recurs throughout his career. “I knew I could do it better,” he said, and he grew the operation from a couple of hundred titles a month to 10,000 before walking away from it with relief, having concluded that the economics were structurally hostile.
“The publishing business is an awful business,” he said. The lesson endured because it was not about books but about position. “If you are a content creator, you have value,” he said. “If you are simply the person in the middle moving things around, it becomes very hard.”
That distinction underwrites everything he has built since, because Riverbed’s entire purpose is to be the thing that holds value rather than the thing that merely passes it along, and across every venture he has applied the same three-part test, which he says has never once varied. “I always look at three things,” he said. “What talent do I have to work with, what technology can I leverage, and what permission do I have in the marketplace to sell something new.” He illustrates the final point with a carmaker that has no licence to sell water, however competent it might be at building cars, and the conclusion he draws is unsentimental. “You have to understand what the market will accept from you next,” he said.
At Riverbed the puzzle is the autonomous enterprise
He joined Riverbed in 2022 and now leads its product management, engineering and data science teams, although he wears the tenure with deliberate lightness given the depth of technology he has inherited. “I am still the new kid,” he said, describing a back catalogue of monitoring technology he keeps returning to. “I walk around and they open up a vault of technology that turns out to solve problems nobody knew they had.”
The organising principle, as ever, is diagnosis in the service of an IT function that is permanently overstretched, and he frames the entire mission as a single recurring question. “How do I make these people more effective?” he said, tracing three generations of artificial intelligence, from reducing the volume of problems reaching IT staff, to offering them informed suggestions, to acting on their behalf. “With the current generation, the message to the IT department is to go and do something else, because the system is handling it automatically in the background,” he said.
That is what the company means by zero disruption, a platform that intercepts a fault before anyone experiences it, and the threshold for acting alone is calibrated to risk. “If we can fix it without any of us even knowing about it, we will fix it in the background,” he said, while the harder cases are held back for human judgement. “There is governance, so we keep a person in the loop to decide whether to act now or hold off,” he said.
His confidence that the largest technology companies cannot simply replicate Riverbed’s position rests on two decades of endpoint and network data, and on an engineering discipline he lays out without embellishment, because he regards the brute-force approach to artificial intelligence as both costly and unreliable. “If you simply throw a language model at the raw data, you are only about 75% right, and you get a different answer each time you run it,” he said. Riverbed instead narrows the data through proven deterministic models before applying its agentic layer. “We reduce the data down deterministically, then bring in an agentic model with skills, zero in on the actual problem, and keep our token consumption to a minimum,” he said.
Asked how he confronts any problem at all, stripped of its setting, he returned to the principle beneath the entire career. “The hardest thing is taking the unknowns and making them known,” he said. “Most of the time, people do not even know what the unknowns are.”
On what comes next he stayed pointedly guarded, unwilling to surrender the advantage of a follow-up conversation. “If I told you all my secrets, why would you want to talk to me next time?” he said, conceding only that several projects he considers genuinely exciting are under way. Then came the advice that closes the loop between a 12-year-old’s misspent lawn money and a lifetime spent fixing things for a living. “Stick to it,” he said. “It is never easy going.”
