Gokul Rajaram: How to Analyse for Durability and Defensibility in a World of AI

TL;DR

Gokul Rajaram presents an eight-mode framework to score software companies' defensibility as AI commoditizes code and threatens traditional SaaS moats.

Key Points

• 1.The Eight Modes framework scores company defensibility. Data, workflow, regulatory, distribution, ecosystem, network, physical infrastructure, and scale — score one point per mode; four or more means you're secure, one or fewer means you're in trouble.
• 2.Data mode requires truly proprietary data. Spotify's Discover product uses a decade of listening behavior across hundreds of billions of users — impossible to replicate cheaply or quickly.
• 3.Workflow mode alone is insufficient. NetSuite (deep ERP) scores a full point; Zendesk scores only 0.5 because lighter workflow embedding is easier to displace.
• 4.Atlassian scores roughly three modes; Monday scores one. Atlassian has proprietary data, workflow, and ecosystem modes — making it massively oversold — while Monday has only workflow, justifying its lower valuation.
• 5.Pure software companies cannot claim scale mode anymore. AI means everyone can now produce software as cheaply as anyone else; only hyperscalers and physical-world companies retain true scale advantages.
• 6.Bolt-on AI strategies have a ceiling unless the entire product experience is rebuilt. Companies must re-evaluate every interaction — e.g., instantly extracting insights from documents upon upload rather than just adding an AI search button.
• 7.BPO spend is the first labor budget AI will capture. Companies cut offshore call centers (Goldman Sachs and Barclays each have 30,000 people in India) before replacing FTEs — the sequence is BPO cuts, then attrition, then layoffs.
• 8.Seat pricing isn't dead but work-output products need outcome-based pricing. Access products like ChatGPT Enterprise keep seat pricing for predictability; work products like Harvey should charge per contract processed, not per user.
• 9.Vertical SaaS must own the full stack to reach $10B+. ServiceTitan went public with 32 products and is still sub-$10B; horizontal companies like Robinhood and Coinbase each have 13–12 product lines doing over $100M revenue.
• 10.Durability metrics matter more than growth rate. Gross retention and net revenue retention are the fundamental quality indicators — Jasper grew from $1M to ~$100M ARR then collapsed, illustrating how raw growth without retention is meaningless.
• 11.Biggest market-sizing miss was Shopify. Rajaram underestimated Shopify at ~$1B valuation because he modeled only existing e-commerce merchants, missing that Shopify would enable any person on the planet to become a seller.