Startups are currently facing unprecedented pressure to achieve rapid growth. To accomplish this, many are striving to secure large enterprises as clients, tapping into the substantial financial resources of the IT sector. However, as any organization engaged in enterprise sales will attest, acquiring such contracts is a formidable challenge. A common misstep among emerging startups is the assumption that their technology will essentially sell itself, according to Arsalan Tavakoli-Shiraji, co-founder and senior vice president of field engineering at Databricks, a company that specializes in data storage and analysis. While effective technology is essential, he cautions that it is insufficient on its own.
Databricks has evolved into a powerhouse in AI and big data storage, boasting a valuation of $190 billion and serving over 70% of the Fortune 500. However, it began as a small startup, and Tavakoli-Shiraji played a pivotal role in securing its initial significant contracts. “When startups sell to enterprises, we often believe the focus is solely on the technology and having the best product,” Tavakoli-Shiraji remarked during his address at TechCrunch Disrupt last week. He learned through experience that “the most critical factors are the people and the processes involved.” He recalls an early encounter with a potential client, only to find out that this individual had already committed to a substantial deal with a major competitor. “He was not interested in hearing what I had to say because he had staked his career on that competitor,” Tavakoli-Shiraji explained. He also reminisced about initial deals falling through due to the startup’s unpreparedness for legal, contractual, or security inquiries.
He offered several key strategies for startups aiming to attract large customers:
Leverage a larger tech company’s network
In its early days, Databricks struggled to secure a meeting with UBS, the prominent Swiss bank. The situation changed when it formed a partnership with Microsoft, becoming an Azure cloud partner. This collaboration meant that sales agreements would be governed by Microsoft’s terms, and crucially, Microsoft’s sales team facilitated the meetings and advocated for the startup. “Suddenly, obtaining that meeting was no longer a challenge,” the founder recalled. “Partnering with an organization that had such access and distribution was immensely beneficial for us.” With the support of Microsoft’s sales personnel, “we began to receive inquiries from all directions.” This influx created new challenges, such as prospects requesting features for regions where Databricks had yet to establish a presence, but these were considered positive developments. Today, Databricks has become a major player that collaborates with startups, marketing their products under its own terms and cybersecurity framework. Notable examples include Hightouch, which offers AI-driven marketing and customer data solutions (in which Databricks has also invested); Addepar, which provides data tools for investment professionals; and Tecton, a machine learning startup that Databricks later acquired.
Address the “four Cs” when selling AI
Tavakoli-Shiraji categorizes the primary concerns and objections of enterprises into four groups he refers to as the four Cs: context, choice, cost, and control. Context involves an AI system that can be trained similarly to onboarding a new employee, adapting to the company’s terminology, processes, and priorities. Choice refers to the ability to switch between AI models rather than being locked into a single provider. Cost pertains to managing the operational expenses associated with the AI product. Control encompasses the measures that regulate the AI’s behavior and safeguard sensitive information.
Challenge the expensive legacy systems enterprises believe they must retain
Tavakoli-Shiraji humorously inquires of CIOs where they perceive their AI capabilities to be on a scale from one to five. A rating of one indicates that employees merely have access to ChatGPT, while a five signifies that the organization has integrated new AI capabilities that have transformed its business processes. “Everyone tends to choose ‘I’m three leaning towards four.’ However, 99% of those I engage with fall into the first or second category.” Enterprises have yet to determine how to deploy AI agents in ways that generate value beyond enabling engineering teams to write code. “It will take many years before organizations can effectively implement AI in a truly meaningful manner,” he stated. A significant opportunity for startups lies in modernizing costly legacy systems. “If you visit any bank today, they still utilize [an IBM] AS/400 and mainframes,” he noted. “No one wanted to engage with it because the functionality of that code was unclear.” However, contemporary AI agents “can analyze and deconstruct what that code does.” This capability, he asserts, paves the way for products that can either enhance the user experience of these outdated systems or replace them entirely. Tavakoli-Shiraji also shared his forecasts regarding the future of enterprise AI adoption, along with a story from the company’s early days when he impersonated a lawyer to secure a significant enterprise contract. See the full interview below.
