Databricks Co-Founder Lays Out How Small Startups Can Win Big Corporate Clients

Long before Databricks became a multibillion-dollar enterprise, its founders were learning the hard lessons of selling to large corporations. Arsalan Tavakoli-Shiraji, now senior vice president of field engineering at the data-and-AI company, told a crowd at TechCrunch Disrupt that founders often underestimate the nontechnical hurdles in landing major accounts. Databricks today is valued at roughly $190 billion and counts more than seven out of 10 companies on the Fortune 500 list as customers, but in the early years Tavakoli-Shiraji led the effort to convert skeptical buyers into long-term partners.

He says many young businesses assume superior code alone will open doors. In practice, relationships, contractual readiness and internal processes determine whether a pilot ever turns into a signed deal. He recalls early meetings that collapsed because the prospect had already staked their reputation on a rival vendor, and others that stalled when the startup couldn’t answer legal or security queries. The takeaway: have your commercial, legal and security playbooks in place before you start courting enterprise accounts.

One fast track, he argues, is to attach your offering to a larger vendor’s sales motion. In Databricks’ case, partnering with a major cloud provider meant the partner’s contract terms and salesforce made introductions that were otherwise impossible to secure. That arrangement also brought a surge of interest — and new headaches like requests for features in regions the startup hadn’t yet supported — but Tavakoli-Shiraji views those as preferable to no meetings at all. Today Databricks plays the reverse role, allowing smaller firms to sell under its compliance controls and procurement terms, a model that has helped a number of niche players grow quickly.

When it comes to AI sales, he groups customer worries into four practical buckets he says startups must address: understanding company context, keeping choices among models open, controlling operational costs, and ensuring governance and data protections. In plain terms, buyers want an AI that learns the firm’s language and workflows, the ability to swap underlying models, transparency on compute and licensing bills, and hard guarantees that sensitive information won’t leak. Startups that can present service-level agreements, model-governance policies and cost-monitoring dashboards will win more trust.

Another clear opening for younger companies is attacking legacy systems that organizations believe are immovable. Many banks and insurers still run business-critical logic on outdated mainframes and bespoke codebases. Today’s automated AI tools can inspect legacy applications, map what they do and speed up migration or create modern user layers on top of them. Industry advisers say focusing on that “retire-and-replace” or “wrap-and-modernize” work can deliver measurable savings fast, a narrative that resonates with CFOs more than abstract model benchmarks.

Tavakoli-Shiraji rounded out his advice with a reminder that sales is a people business: staffing customer-success, legal and security teams matters as much as product polish. He even shared a cheeky early-career anecdote about taking unconventional steps to move a negotiation forward — which, he noted, underscored the importance of having actual counsel and compliance processes in place as you scale. For startups chasing their first large customer, the simple plan is clear: be technically solid, but build the human and operational scaffolding enterprises require.