The Futurist

Run Python & R code inside your search sessions, sandboxed and persistent

“The thesis here is falsifiable: retrieval and computation will converge into a single interface, and the tool that owns the retrieval layer will own the compute layer by extension, because users won't tolerate the context switch. The dependency that has to hold is that Perplexity retains a meaningful share of the search-for-research workflow against both Google's AI Overviews and ChatGPT's browse-plus-analyze combo — that's a real bet, not a given. The second-order effect that nobody's talking about: if this pattern works, it reframes what a search session is. Right now search is read-only; adding a persistent stateful compute environment makes it read-write, which changes how researchers, analysts, and journalists interact with live information. The trend line is the collapse of the research-to-analysis pipeline into a single context, and Perplexity is on-time to it — not early, but not late enough to be irrelevant. The future state where this is infrastructure is when 'search and analyze' is a single verb and Perplexity is the default runtime for it.”

RAG model with citation-level grounding for regulated enterprise search

“The thesis is falsifiable: within three years, enterprise AI adoption in regulated industries will be gated on auditability at the response level, not just model-level safety filters, and organizations will pay a premium for models where every claim traces to a source document. The second-order effect that's underappreciated here is what citation-grounded RAG does to knowledge work accountability — when the AI's answer includes a source link, the human reviewer shifts from 'is this true' to 'is this source authoritative,' which is a fundamentally different cognitive job and changes how knowledge workers are trained and evaluated. Cohere is riding the trend of enterprise AI deployment moving from experimentation to compliance-gated production, and they're on-time to early — most regulated-industry AI deployments are still in pilot phase. The dependency that has to hold: enterprises must continue to face regulatory pressure that makes 'the model said so' an insufficient answer, which every current signal in financial services and healthcare regulation suggests will intensify, not relax.”

Extended thinking for grad-level math, science, and coding

“The thesis o3 Pro is betting on: that inference-time compute scaling is a durable lever for capability gains, and that users will pay a premium for correctness on high-stakes problems rather than just throughput. The dependency that has to hold is that extended thinking produces calibrated confidence improvements, not just longer outputs that feel more authoritative — the research trend on compute-optimal inference scaling broadly supports this but is not settled. The second-order effect that matters here is the shift in who gets access to expert-grade reasoning: a researcher at an institution without a PhD supervisor can now get graduate-level feedback on their methodology. That's not marginal, that's a structural redistribution of intellectual leverage. OpenAI is on-time to the inference scaling trend — not early, not late — and o3 Pro is the right shape of product for it. The future state where this is infrastructure is one where extended thinking is the default mode for any query touching scientific or engineering decisions.”