AI & technology
Agentic & autonomous AI
14 indexed claims · 7 guests · peaked 2025-Q3
Overview
Agentic AI — systems that act on behalf of users without step-by-step instruction — is the topic generating both the most excitement and the sharpest warnings across these podcast episodes. Michael Nevski sees it enabling a real-time, deterministic picture of consumer behavior that periodic surveys never achieved. Jason Cohen reports it already collapsing hours of data preparation into minutes on the research workbench. But the same capability that empowers researchers introduces new threat surfaces: AI agents completing surveys undetected, automation shipping the wrong message to millions before anyone catches it, and systems granted decision-making authority before the industry has governance standards to take it back. The editorial center of gravity is not whether agentic AI will transform research — that debate is largely settled — but whether the field is building the infrastructure to capture the upside without absorbing preventable damage.
From the corpus
- Rolling out agentic AI marketing workflows at scale without a validated, real-time representation of the customer is genuinely dangerous — the same automation power that enables personalization can ship the wrong message to the wrong segment at enormous scale before anyone notices. — Kate O'Keeffe, Ep. 55 ·
cl-kate-okeeffe-016 - Agentic AI's top benefit for research is real-time insights: once consumer trust and privacy protections are in place, researchers will get a complete 360-degree view of the consumer path to purchase across channels — not a periodic survey, but continuous daily data. — Michael Nevski, Ep. 26 ·
cl-michael-nevski-003 - Software, programming, and even Excel skills are no longer the bottleneck for researchers; agentic command-line AI (Claude Code, Codex) means data cleaning and analysis can be driven by natural-language vision rather than personal coding capability. — Jason Cohen, Ep. 43 ·
cl-jason-cohen-027 - Deterministic data — built directly from actual transactions rather than modeled or look-alike estimates — is the gold standard for behavioral insights, and agentic AI is the bridge that will make it possible to build this across the ecosystem. — Michael Nevski, Ep. 26 ·
cl-michael-nevski-015 - Letting Claude Code orchestrate verified internal data-cleaning utilities collapses hours of manual data prep into a couple of minutes; the human's job shifts from doing the cleaning to picking which trusted utilities the AI will run. — Jason Cohen, Ep. 43 ·
cl-jason-cohen-029 - AI agents completing surveys is already happening and represents the next-generation fraud vector — the behavioral signals needed to detect agent-based fraud are categorically different from those that catch human fraudsters. — Roddy Knowles, Ep. 8 ·
cl-roddy-knowles-011 - Agentic AI usage is already standard at the big tech firms, but is unevenly distributed across the broader industry; the lag in adoption — not the technology itself — is what creates the present opportunity (and risk). — Jason Cohen, Ep. 43 ·
cl-jason-cohen-028 - Hyper-personalization enabled by consumer opt-in to transactional history tracking is already a live capability at scale — not a future concept — and agentic AI will make it possible to deliver those experiences in real time without manual intervention. — Michael Nevski, Ep. 26 ·
cl-michael-nevski-019
Contributing guests
Episodes in Agentic & autonomous AI
- Closing the say-do gap: what live experiments reveal that surveys miss with Kate O'Keeffe — Ep. 55 · Apr 21, 2026
- How causal AI is revolutionizing consumer and market research with Jason Cohen — Ep. 43 · Jan 20, 2026
- The Gig Economy, Affluent Consumers, and What They Mean for Brands — Ep. 26 · Sep 16, 2025
- The Real Cost of Bad Data: Survey Fraud, AI Agents, and the Fight for Data Integrity — Ep. 8 · May 13, 2025