Great technology won’t close an enterprise deal on its own. That’s the hard lesson from Arsalan Tavakoli-Shiraji, co-founder and senior vice president of field engineering at Databricks, who shared his sales playbook at TechCrunch Disrupt.
Quick Facts
- Speaker: Arsalan Tavakoli-Shiraji, Databricks co-founder
- Company: Databricks, valued at $190 billion, with over 70% of the Fortune 500 as customers
- Core lesson: People and processes matter more than the best tech
- Source: TechCrunch (reporter Julie Bort, October 26, 2026)
Why Great Tech Isn’t Enough
Young startups often assume a superior product will sell itself. Tavakoli-Shiraji says he learned otherwise while landing Databricks’ first big deals. Early on, he once landed a meeting with a prospect who had already staked his career on a rival’s mega-contract, so he had no reason to listen.
Other early deals fell apart because the startup wasn’t ready for legal, contract and security questions. Enterprise buyers expect answers to all three before they sign.
Tip 1: Sell Under a Bigger Company’s Umbrella
Databricks tried and failed to get a meeting with Swiss bank UBS. Then it became an Azure cloud partner with Microsoft. That meant deals could run under Microsoft’s terms, and Microsoft’s sales team booked meetings and advocated for the startup.
Suddenly, meetings that had been hard to get became easy. Tavakoli-Shiraji said the startup started “getting pulled everywhere,” which created new problems, like prospects asking for features in countries Databricks hadn’t entered yet. Those were good problems to have.
Today, Databricks plays the same role for startups. It sells their products under its own terms and security setup. Examples include Hightouch, Addepar and Tecton, which Databricks later acquired.
Tip 2: Answer the “Four C’s” When Selling AI
Tavakoli-Shiraji groups enterprise objections about AI into four categories:
- Context: Can the AI learn company terminology and processes, like a new employee?
- Choice: Can customers switch models instead of locking into one provider?
- Cost: Can they control what the AI costs to run?
- Control: Are there safeguards for behavior and sensitive information?
Address all four early, and you remove the biggest sources of buyer hesitation.
Tip 3: Go After Expensive Legacy Systems
Tavakoli-Shiraji asks CIOs to rate their AI progress from one to five. A one means staff merely have access to ChatGPT. A five means AI has reshaped business processes. Nearly everyone claims “three leaning toward four,” but he says 99% of the leaders he talks to sit in the one-or-two range.
He believes real enterprise AI adoption will take many years. In the meantime, he sees a big opening in legacy tech. Banks still run IBM AS/400s and mainframes that nobody wants to touch because nobody knows what the code does. AI agents can now analyze and break down that code, which creates openings for startups to modernize interfaces or replace the systems entirely.
Takeaways for Founders
- Treat sales process as a product: legal, contracts and security matter.
- Find a larger partner whose sales team can open doors.
- Frame your AI pitch around context, choice, cost and control.
- Look for painful, expensive legacy systems that enterprises avoid touching.
What to Watch Next
Expect more startups to pursue partner-led sales and legacy-modernization products as AI agents mature. TechCrunch’s full interview also covers Tavakoli-Shiraji’s predictions for enterprise AI, plus a story about posing as a lawyer to land a deal.
Source: TechCrunch