Startups across the technology sector are currently facing unprecedented levels of pressure to scale rapidly and achieve sustainable growth. To meet these aggressive targets, many young companies realize they must eventually land large enterprise organizations—the deep-pocketed behemoths of the global IT world—as primary customers. However, as any veteran B2B software executive will readily confirm, navigating the complex corporate procurement landscape and winning these high-stakes enterprise contracts is rarely a straightforward or easy endeavor.
According to Arsalan Tavakoli-Shiraji, co-founder and senior vice president of field engineering at Databricks, one of the most critical and recurring mistakes made by early-stage founders is the naive belief that their cutting-edge technology will practically sell itself. While acknowledging that product excellence is a fundamental prerequisite, Tavakoli-Shiraji warns that having superior software alone is simply not enough to close enterprise deals.
Today, Databricks stands as a towering giant in the artificial intelligence and big data storage sectors, boasting a massive valuation of $190 billion and counting more than 70% of the Fortune 500 among its active customer base. But the enterprise titan was once a modest startup fighting for survival, and Tavakoli-Shiraji was the person directly responsible for helping secure its first major breakthrough agreements. Reflecting on those formative years, he shared valuable lessons with attendees at TechCrunch Disrupt, emphasizing that while the code matters, human elements and corporate processes ultimately dictate success.
"Everybody when they’re selling to enterprises at startups… we think it’s all about the tech and having the best tech," Tavakoli-Shiraji told the audience. Over the years, however, he learned the hard way that people and processes matter the most in enterprise software sales. He recalled an early-stage attempt to secure a meeting with a high-value prospect, only to discover upon arrival that the prospective buyer was the exact individual who had recently signed a massive, career-defining mega-deal with Databricks’ primary market competitor. The executive had virtually no interest in listening to an alternative pitch because he had essentially bet his professional reputation on the competing product.
On other occasions, early deal pipelines fizzled out during late-stage evaluations because the fledgling startup was completely unprepared to answer rigorous legal, contractual, or security inquiries from corporate risk management teams. Drawing from these hard-earned experiences, Tavakoli-Shiraji outlined core strategies and perspectives for startups attempting to break through the enterprise barrier.
Sell Under a Bigger Tech Company’s Umbrella
During the early days of Databricks, the company made repeated, unsuccessful attempts to secure a meeting with UBS, the prominent Swiss banking institution. The doors remained firmly closed until the startup executed a strategic partnership with Microsoft, becoming an official Azure cloud ecosystem partner. That alliance meant that Databricks’ sales agreements would be covered under Microsoft’s established enterprise terms, providing an immediate layer of corporate trust and legitimacy.
More importantly, the partnership meant that Microsoft’s dedicated sales representatives actively booked meetings and advocated for the startup’s technology within their own extensive client networks. "All of a sudden, that meeting that wasn’t hard to get," the co-founder recalled, noting the dramatic shift in market momentum. "It was a huge boon to us that we were able to partner with somebody who had that access and distribution."
With the immense weight of Microsoft’s sales force helping drive market awareness, Databricks began experiencing a massive pull across multiple industries. That influx of interest naturally generated new operational challenges, such as prospective clients demanding localized product features for geographic regions where Databricks had not yet expanded. Nonetheless, management viewed these obstacles as positive growing pains stemming from accelerated market penetration.
In a full-circle evolution, Databricks has now grown into the major enterprise player that actively works with smaller startups, offering to distribute and sell their specialized products under Databricks’ own established legal terms and comprehensive cybersecurity frameworks. Notable examples of this modern ecosystem strategy include Hightouch, a company that provides AI-powered marketing and customer data tools—and in which Databricks has made direct strategic investments—alongside Addepar, a provider of specialized data analytics tools for investment professionals, and Tecton, a machine learning startup that Databricks eventually acquired.
When Selling AI, Address the "Four C’s"
When discussing the complexities of marketing artificial intelligence solutions to risk-averse corporate buyers, Tavakoli-Shiraji categorizes enterprise concerns and operational objections into four distinct categories, which he terms the "four Cs": context, choice, cost, and control.
Context refers to the critical need for enterprise-grade AI systems to be trained effectively—a process he likens to onboarding a new human employee. The technology must thoroughly understand an organization’s proprietary terminology, internal workflows, strategic priorities, and operational nuances before it can deliver genuine business value.
Choice represents the enterprise buyer’s insistence on maintaining the commercial freedom to switch between different AI foundation models rather than being locked into a single proprietary vendor ecosystem. Cost addresses the pressing requirement for transparent financial governance, giving corporate leaders the tools to monitor and control the computational expenses required to keep the AI product running at scale. Finally, control encompasses the robust security safeguards and governance frameworks necessary to govern automated AI behaviors and protect sensitive proprietary information from breaches or compliance violations.
Attack the Expensive Legacy Systems Enterprises Think They’re Stuck With
When evaluating the current state of enterprise AI adoption, Tavakoli-Shiraji often lightheartedly asks chief information officers to rate their organization’s AI maturity on a scale from one to five. A rating of one signifies a basic environment where employees merely have access to consumer tools like ChatGPT, while a rating of five indicates that the enterprise has fully integrated transformative AI capabilities capable of fundamentally reshaping its core business processes.
Although many corporate leaders jokingly claim to be a three leaning toward a four, Tavakoli-Shiraji estimates that roughly 99% of the executives he speaks with actually reside firmly in the first or second category. Most enterprises have yet to figure out how to deploy autonomous AI agents in ways that generate measurable business value beyond simply assisting engineering teams with writing software code. Consequently, he suggests that it will take many years before organizations can properly and pervasively deploy AI in a truly meaningful capacity across traditional enterprise workflows.
For agile technology startups, this widespread hesitation presents a massive commercial opportunity centered on modernizing expensive, entrenched legacy systems that corporations assume they are permanently stuck maintaining. Pointing to the financial sector, Tavakoli-Shiraji noted that nearly every major traditional bank still relies on legacy architecture such as IBM AS/400 systems and ancient mainframes. Historically, internal IT departments avoided touching these aging systems because nobody left within the organization understood what the underlying code actually did.
Today, however, advanced AI agents possess the capability to systematically examine, interpret, and decompose what legacy code actually accomplishes. According to Tavakoli-Shiraji, this technical breakthrough opens the door for innovative startups to build products capable of everything from wrapping modern user interfaces around ancient infrastructure to replacing outdated core systems altogether.
