Elias Torres on Selling Drift, Hiring and Building Again with AI

Elias Torres calls Drift, the company he co-founded and sold for $1.2 billion, the biggest failure of his life. His standard is personal: he wanted an enduring company and product, and regrets not taking it further. In this March 2025 interview, the Agency co-founder and CEO talks with Harry Stebbings about immigration, AI, hiring and family.

Torres describes saying “I do not care” as a way to cope with caring intensely. Years of rejection, lost customers and employees leaving have taught him that setbacks can be survived. Financial security, he adds, removes many everyday material worries.

Stebbings worries that reassuring a young founder that everything will be fine removes the fear that drives achievement. Torres would instead tell his younger self that many things seem harder than they are, while distinguishing starting a company from successfully finishing and exiting one.

Growing up with little in Nicaragua shapes his perspective on hardship and complaints about workplace perks. He thinks immigrants often develop grit through difficult experiences, while acknowledging that determination is not confined to people who grew up poor.

Torres recalls losing the family home after its owners returned following Nicaragua’s political upheaval. His grandmother’s immigration history helped the family obtain green cards, and they moved to the United States when he was finishing school. He saw the move as an opportunity.

Earlier, during Nicaragua’s civil war, his mother sent him to Los Angeles to live with his father and avoid the draft. There he first used an IBM computer, typing homework and experimenting with software; the device sparked an attraction to technology.

Asked whether wealthy children have advantages, Torres points to access, networks and money his own children will have. He recalls his daughter challenging his advice to quit a job without worrying about the bonus: she told him not everyone has the financial freedom he now enjoys.

He says he tries to create adversity for his children, while acknowledging that his own success arrived after they were older. He then recalls joining a special IBM internet technology group and building browser applications with HTML and JavaScript in the late 1990s.

Working inside a huge company taught Torres that size limits speed. Stebbings challenges him with Adobe and Google as examples of incumbents moving quickly on AI; Torres disagrees and argues that public announcements can exaggerate the pace of meaningful product progress.

Torres criticizes Google and Microsoft’s AI execution, while praising Microsoft’s early investment in OpenAI. His assessments of specific products and their adoption are personal, time-bound opinions in the interview, not independently established conclusions.

Stebbings argues that Google’s consumer distribution gives it enormous potential. Torres agrees distribution and data are powerful incumbent advantages, but says that is a separate question from whether large companies have moved quickly. Stebbings likewise expects Microsoft’s distribution to help adoption despite reservations about Copilot.

For Torres, speed creates openings for challengers. He sees today’s chat interfaces as a step toward, rather than the endpoint of, AI software: they still require a human operator. His analogy is the difference between a car designed around a driver and one that can carry out the journey itself.

He blames organizational friction for slower adoption: approvals, management layers, internal reporting and employees whose incentives reward avoiding risk. A public-company CEO’s time with analysts and the press can also reduce attention to customers and products, in his view.

Torres dislikes routine one-on-one meetings when they substitute for working together. At the time, he says, Agency had roughly 15 people and one team meeting focused on what each person did and will do. As teams grow, his preferred answer is small groups owning distinct products.

He predicts that AI could allow companies to earn billions in revenue with much smaller teams. Customer success is his example: at Drift, he says, it was difficult to hire and train enough people to support customers. His new company Agency sells AI for customer success, so it has a direct commercial interest in this thesis.

He adds that this outcome is a possibility, not a certainty. Executives may endorse a product but delegate the purchase through layers that resist change, he says; he urges CEOs to own the decisions they say matter. That account also reflects his experience selling Agency to enterprises.

Stebbings asks whether smaller teams threaten venture capital. Torres suggests lower staff needs could be offset by compute spending and agrees that financing might shift toward less equity and more debt against predictable revenues. Both describe possibilities rather than established industry outcomes.

Torres argues that business software was built around human operators and predicts AI will change that assumption. He disputes the idea that companies will necessarily keep expanding headcount, suggesting people may work at different companies or start businesses serving neglected markets.

He is optimistic that people can pursue other work and interests, including farming, travel and art. This is his outlook on a difficult labor-market transition, not evidence that every displaced worker will readily find an equivalent opportunity.

On recruiting, Torres looks for initiative, intelligence, grit, speed and obsession. He says his interviews are intense: he asks candidates to show code and explain exactly what they built. His aim is to find people who want to work in that demanding environment.

Stebbings asks whether a more adaptable leadership style would get the best from different people. Torres resists changing himself to suit everyone and connects his past attempts to accommodate a larger team with his frustration at Drift. The claim that a different approach would have produced a public company is his counterfactual judgment.

Torres calls Drift his biggest failure because he wanted to build a durable product and felt he fell short for customers. He blames himself for insufficient product improvement and execution speed, and thinks the company had become too large and inefficient to adapt easily.

He says accepting a sale was a pragmatic decision despite comparisons with peers receiving higher valuations in 2021. Drift was sold to Vista for $1.2 billion, as described in the episode. He declines to disclose his personal proceeds but says the outcome gave him life-changing financial security.

Money lets him spend less time on material logistics and more on customers and his team. He values assistance and says founders should decide what deserves their attention rather than treating frugality as an end in itself.

His recruiting approach has changed from persuading as many people as possible to pushing back and testing fit. He says candidates can learn to give him the answers he wants, so he now uses paid part-time contract work to see how they actually perform.

He describes the change as hiring slowly and firing quickly. Early founders, he says, may overestimate a hire because they have not yet developed recruiting judgment; he now tries to be clearer about expectations before people join.

Torres recounts hiring five Northeastern students after reviewing their work and connections. He brought new hires to his house, gave them laptops and started shipping code immediately. He says this worked for his small team, though it is one founder’s account rather than a general result.

He believes founders often recognize a poor fit in the first week or two but ignore the signs. His suggested alternatives are deeper references, better testing and paid trial work. Stebbings notes that letting someone go so early can feel unfair or damaging to a founder’s reputation.

Torres also warns against assuming that quiet candidates are weak. He recalls finding overlooked engineers who demonstrated grit and strong code, then sometimes making a hire despite a skeptical interview panel. His stated preference for high-intensity interactions does not mean he dismisses every shy applicant.

His biggest hiring mistakes, he says, came when he saw warning signs but convinced himself to proceed because he needed help. He advises founders facing sales targets or investor pressure not to lower their standards simply to fill a role.

A résumé alone cannot settle the question: short tenures may warrant questions, but long tenure and title progression also require examination. Torres asks candidates what they actually did and is skeptical of titles as a substitute for impact.

He is considering ways to reward small teams through equity, salaries or profit and revenue sharing rather than titles alone. These are ideas he is exploring for Agency, not a finalized compensation plan or a promise of future returns.

Torres says he entered startups for the work of building, serving customers and creating with a team. On profit sharing, he says the company must first reach profitability and decide how it will be built.

He now pays more attention to profitability and free cash flow than he did when cheap capital made growth the dominant discussion. He wants to build an enduring company, but agrees with his former collaborator David Cancel that founders often have to balance growth and profitability simultaneously.

Stebbings worries that extremely fast-growing AI companies will raise investor expectations and leave slower enterprise software businesses unfunded. Torres shares the concern and says Agency’s enterprise model requires trust and customer relationships, which may take longer to build.

Asked about firing, he says he tries not to make promises he cannot keep and to be upfront about the intensity of the work. He enjoys closing deals, including hiring, sales and fundraising, but values the relationships that continue afterward.

Selling Drift had an emotional cost. Torres says new leadership left him feeling less relevant in the company he created; he spent roughly seven to ten months traveling and finding his footing afterward. He sees ego as both a danger and a source of motivation.

Torres advocates having children younger based on his own experience and concerns about fertility, while acknowledging that not everyone wants children. These are personal views, not medical advice. He says he met his wife at 17 and had the first of their three children at 27.

He credits his wife, who was at home with their children, and describes the exhaustion of working at IBM, parenting, commuting and doing chores without nearby family support. Stebbings asks whether those demands compromise a founder’s performance; Torres says his own family and work coexisted, although the experience was difficult.

He remembers handling startup outages and taking a laptop to the beach instead of enjoying weekends or travel. He feels fortunate that, after those trade-offs, he can now spend time and travel with his grown children; he does not claim the same path is right for everyone.

On marriage, Torres jokes that his wife is always right, then says a lasting relationship requires balancing one’s own desires with those of others. He finds pleasure in making room for what his family wants as well as what he wants.

As a father he has learned to show affection and offer the words of affirmation his children ask for. He also believes in building resilience. He recalls harsh treatment in his own childhood while describing his mother with love; his reflections are his own account of parenting.

Torres often worried he was letting his children down because work left little time or money for tutors and activities. Stebbings recognizes the feeling across his own roles. Torres says nobody can perform every role perfectly, and tries to focus on the good he can do rather than satisfying every expectation.

Returning to Drift, he compares his ambition with HubSpot’s continuing founders and with Klaviyo. Torres says he was an early investor in Klaviyo, encouraged its founder to take venture funding and still owned stock at the time; his praise of the company should be read alongside that financial interest.

He considered becoming a venture capitalist but decided he preferred building to advising other founders and raising a fund. He says he earned about $2 million over six or seven months doing AI consulting, including work for OpenAI customers, which helped him learn and eventually informed Agency’s idea.

Torres predicts AI services companies could become larger than AI technology providers over the following five years because organizations need help adopting the technology. He expects the transition to take longer than one or two years and cautions that workflows and solutions were still developing.

In a quick-fire round, he champions directness and chooses Nvidia when asked for a public stock to hold for ten years, citing demand for compute. He also names Salesforce as his short; these are personal market opinions at the time, not investment recommendations.

He says he changed his mind about founding another venture-backed company and now wants to give his children freedom from his expectations. Money makes life easier, he says, but does not itself create happiness. Extremism and the inability to have a conversation concern him.

After visiting Egypt, Torres says he is inspired by historic craftsmanship and hopes technology can help people build better products, customer relationships and businesses, while making more room for art, culture and family. He ends on that optimistic aspiration.