Key insights
- The post questions the economic viability of replacing SaaS solutions with AI, specifically focusing on the high token costs associated with AI models. It highlights the potential for traditional SaaS to remain more cost-effective for enterprise functions like HR, suggesting a bearish outlook for AI's immediate disruption of the SaaS market.
Not really a tech guy, so I’m hoping someone familiar with enterprise AI cost structures can help me out.
I’m currently on Claude’s $200 plan, mainly for coding and research. I regularly hit the token limit and have to wait 4 hours for the next session to reset.
Let’s say an HR department is currently running 3–4 separate software tools (ERP, payroll, timesheets, etc.). How many tokens would it roughly take for AI to fully handle those same tasks?
At what point does the cost of using AI become so much higher than running standalone SaaS tools that the traditional software ends up being the cheaper (and better) choice? In other words, will AI actually be able to replace SaaS in this kind of scenario, or will the token costs make it impractical?