Can AI-native sales stacks disrupt Salesforce's $200B? Owning the sale, not the CRM
CCgong · x · 2026-09-18
A long analysis posted while stuck in Dreamforce traffic argues Salesforce ($200B market cap, 27 years old) is vulnerable to a wave of AI CRMs with different flavors: rip-and-replace (Attio), full-stack sales OS (MonacoGTM), and agents layered on existing CRMs (rox.ai).
The core thesis: the opportunity isn't a better CRM, it's owning the sale. Deals increasingly live across texts, in-person events, email, calls, Slack and contracts—whoever captures the traces of decisions leading to a closed deal builds a learning loop Claude can't reconstruct from Salesforce's fields. Sales is a great learning loop: high-frequency work where judgment matters, so a good AE can grade an agent's output in seconds.
Salesforce's schema is a tax for relationship-driven jobs (recruiting, VC, real estate, agencies), and seed-to-Series-B companies can't be served profitably with a Salesforce implementation—which is why forward-deployed seller + software models like Monaco's are landing. Its 150k customers won't churn tomorrow, but the next 10k AI-native companies running on 5 humans and 50 agents will never install it. As for companies building their own CRMs: they end up building a schema, which was never the hard part—the hard part is integrations that break weekly, licensed external data, deliverability, and a learning loop sharpened by ten thousand teams correcting the same agent.
More from Companies & People
- Sakana AI launches Frontier Intelligence Group to pursue alternatives beyond the Transformer paradigm — SakanaAILabs · 2026-09-18
- Nat Friedman's Muse reportedly keeps nearly every interaction under 150ms — armand_ruiz · 2026-09-18
- Sakana AI's Takuya Akiba and CADDi's Yoshihiro Fukuhara on How to Compete Globally in AI Research — tkasasagi · 2026-09-18
- Huawei's Eric Xu: Six or Seven Chinese AI Labs Target 10T-40T Parameter Models in Two Years — EleanorOlcott · 2026-09-18
- Why don't LLM labs compete on personality? Users value it over raw capability — dioscuri · 2026-09-18
- Harvard's free ML systems curriculum goes full AI engineering, 28K stars on GitHub — techNmak · 2026-09-18