AI, Growth, and the Future of Healthcare | Anish Acharya & Sachin Jain
Andreessen Horowitz General Partner Anish Acharya speaks to SCAN Health Plan leadership about AI adoption, arguing that AI represents the most transformative technology since the wheel. He outlines three key areas of AI deployment—chat, coding, and customer support—while emphasizing that healthcare's 45% administrative cost burden makes it the most important sector for AI-driven efficiency gains.
Summary
In this fireside chat, A16Z General Partner Anish Acharya addresses SCAN Health Plan's leadership team, framing AI as the most significant technological development in human history, potentially since the invention of the wheel. He contextualizes AI within broader patterns of technological disruption, noting that technologies like electricity historically transferred 80% of created value to consumers, turning luxuries into commodities—a pattern he believes AI will replicate at unprecedented scale.
Acharya identifies two camps of AI perception: the 'Instagram view' held by those who tried ChatGPT early and remain unimpressed, and the 'Twitter view' of deeply convinced believers experiencing what he calls 'AI psychosis.' He argues the truth lies between these extremes but insists that skeptics simply haven't engaged with the tools recently enough, as capabilities are compounding every three to six months.
On practical deployment, Acharya outlines three working areas: chat as a thinking partner (not just an email drafting tool), code generation that he argues should be democratized to every employee regardless of technical background, and customer support agents that he believes are widely misunderstood. He pushes back against the narrative that AI customer support is purely a cost-cutting measure, arguing instead that it frees human agents to handle high-value interactions and enables a blurring of sales, support, operations, and collections functions.
Acharya addresses organizational concerns around AI governance, arguing that the existential risk of under-adopting the technology far outweighs the risks of over-exploration. He recommends a hybrid centralized/decentralized model where IT creates infrastructure for experimentation, a central AI team distributes best practices, and individual teams are empowered to pull their own threads. He uses token consumption as a proxy metric for genuine AI engagement, noting that a whole company spending $625/month on AI is insufficient.
On the public market selloff of SaaS stocks, Acharya argues the market has overcorrected, pointing out that most enterprise software touches regulated systems where errors carry serious liability, making vibe-coded replacements implausible. He also suggests financial markets are selling existing names to create capacity for anticipated IPOs from OpenAI, Anthropic, and SpaceX.
For the future of work, Acharya suggests AI will more likely produce a four-day workweek than mass unemployment, reduce corporate politics by introducing impartial AI decision-making, and eliminate administrative overhead that prevents workers from doing their best work. He closes by arguing that making healthcare—which carries 45% administrative cost at the industry level—dramatically more efficient through AI represents the greatest opportunity for broad societal benefit, potentially driving deflation rather than merely disinflation in healthcare costs.
Key Insights
- Acharya argues that AI customer support is not primarily a cost-cutting tool but a function-merger that enables human agents to fluidly combine sales, support, operations, and collections roles while AI handles low-value queries.
- Acharya claims that 45% of healthcare spending is tied to administration rather than direct patient care, making healthcare the highest-stakes sector for AI-driven efficiency and potentially the greatest opportunity for societal benefit.
- Acharya contends that the public market selloff of SaaS stocks is an overcorrection, arguing that most enterprise software is embedded in regulated systems where errors carry legal liability, making AI-generated replacements implausible in the near term.
- Acharya asserts that the risk of under-adopting AI is more existential for an organization than the risks of security incidents or regulatory issues that might arise from aggressive experimentation.
- Acharya argues that AI is fundamentally different from prior technology waves—internet, mobile, electricity—because it can perform work autonomously rather than merely extending human intellectual productivity.
- Acharya claims that AI may reduce corporate politics by replacing human judgment with AI decision-making in tie-breaking situations, making outcomes feel procedurally fair even when individuals disagree with the result.
- Acharya uses monthly token spend per employee as a concrete proxy for genuine AI engagement, citing a portfolio company where the entire workforce spent only $625/month total, which he characterized as a failure of adoption.
- Acharya argues that employees who define their job tasks in reusable AI prompts and distribute them as organizational skills will outperform those who hoard prompts out of job security fears, framing prompt-sharing as an abundance mindset versus a scarcity mindset.
Topics
Full transcript available for MurmurCast members
Sign Up to Access