SaaStr 839: Why Most SaaS Companies Will Fail at AI (And How...

Full Title

SaaStr 839: Why Most SaaS Companies Will Fail at AI (And How to Avoid It) with Intercom's CPO

Summary

This episode discusses the existential challenge AI presents to traditional SaaS companies, arguing that most will fail to adapt without a radical transformation.

The speaker, Intercom's CPO, shares their company's "brutal" journey of pivoting to an AI-first approach, highlighting the necessary cultural shifts, product reimagining, and organizational changes required for survival and success.

Key Points

• Traditional SaaS models of selling "seats" and relying on GUIs for CRUD tasks are becoming obsolete in a post-AI world, where outcomes driven by AI and custom models will determine winners.
• Intercom underwent a significant transformation, betting its future on AI after experiencing declining revenue growth, and launched "Finn," an AI agent that now resolves over a million customer problems weekly.
• True AI companies require a complex, multi-layered approach encompassing AI layers (like RAG systems), model layers (including custom models), and app layers, demanding significant investment and expertise.
• Adapting to AI necessitates a culture of change and the elimination of "sacred cows," meaning companies must be willing to re-found themselves by changing their organization, product, roadmap, metrics, go-to-market strategies, and pricing.
• Building AI products requires a shift from traditional SaaS development, focusing on empirical evaluation, iterative experimentation, and accepting that many product improvements will be invisible and unpredictable, with the UI becoming less critical than the underlying AI infrastructure.
• Companies must avoid the common mistake of merely adding AI features to existing products rather than reimagining them entirely, and they must be willing to make difficult, "self-harming" decisions that might impact short-term revenue for long-term AI-driven success.
• Marketing AI products is challenging due to the prevalence of demos over production-ready products and the difficulty in differentiating visually similar AI solutions, shifting the focus to infrastructure and scientific rigor.
• The buyer landscape for AI products is evolving, involving multiple stakeholders like C-level executives focused on AI transformation and specialized AI personnel, requiring companies to adapt their sales and marketing strategies accordingly.
• Successful AI transformation involves embracing chaos and ambiguity, as the new way of building software starts with understanding what AI makes possible, followed by iterative development and real-world scaling, unlike the stable, execution-focused SaaS era.
• Companies must be cautious of the "marketing overhang," where promises are made based on demos that don't reflect production reality, and must invest in customer education and advocacy because customers are often unsure how to buy or evaluate AI.
• The shift from design being expensive and a bottleneck in SaaS to being cheap and enabling rapid prototyping in AI requires new skill sets for designers, including coding, and a reevaluation of team structures.
• Companies need to identify and remove processes that are no longer relevant in the AI era and be wary of employees who resist change, as they can become liabilities to transformation.

Conclusion

Companies that fail to fundamentally transform into AI-native organizations will likely fail, while those that embrace the change can become category leaders.

The AI revolution is non-negotiable, requiring a complete overhaul of existing business models, product strategies, and company cultures.

Successful adaptation involves rigorous experimentation, a willingness to delete outdated processes, and making difficult decisions, rather than simply adding AI features to existing products.

Discussion Topics

• What are the biggest cultural hurdles your organization faces when considering a significant AI transformation?
• How can traditional SaaS companies effectively differentiate their AI products when many appear visually similar?
• What are the most critical "self-harming" decisions a company might need to make to truly embrace an AI-first future?

Key Terms

CRUD

Stands for Create, Read, Update, Delete; refers to the basic functions of persistent storage in applications.

RAG system

Retrieval-Augmented Generation; a technique that combines retrieval models with generative language models to produce more accurate and contextually relevant outputs.

Timeline