Systems, Not Slides: IBM’s Ray Bahari on AI-Driven Product Marketing and Launches That Learn

Summary

If your launches still run on “plan, ship, celebrate,” you’re leaving revenue on the table.

Ray Bahari, Global Product Marketing Lead at IBM and author of Systems, Not Slides: An AI Blueprint for Product Marketing, lays out how to move from launch theater to launch telemetry—compressing the distance between insight and execution.

Drawing on a career that spans scaling an early e‑commerce company, medtech at Stryker, mobile research at Polefish, insurance growth at Plymouth Rock, brand-building at JPMorgan’s Onyx, and ABM at IBM, Ray shares a practical playbook for adaptive positioning, sales enablement, and brand‑safe content at scale.

He explains how to build “minimum viable loops” across PMM—so positioning, proofs, messaging, and objections continuously inform each other—and how to choose a focused AI stack without drowning in tools.

Expect specifics on variant testing without losing brand consistency, creating a reusable bank of proofs, instrumenting journeys for speed-to-signal, and establishing a test‑and‑learn culture that actually sticks.

Timestamps

[00:45] – Guest intro and career arc: from early e‑commerce to SaaS, medtech, fintech, and IBM

[02:34] – Influencer dynamics in buying: lessons from Oticon and randomized rewards

[04:58] – Inside Stryker: digitizing the field with iPads and measurable customer interactions

[06:54] – Mobile-first research: Polefish and getting 2,000 survey responses overnight

[07:50] – Competing in insurance: partnerships, a homegrown DMP, and incrementality

[09:40] – IBM and JPMorgan: ABM impact, building Onyx, and returning to scale GTM with AI

[11:43] – From theater to telemetry: compression, adaptive positioning, and brand-safe variants

[17:16] – Tool choices, custom GPTs, and building Minimum Viable Loops and culture

Takeaways

- Build systems and loops—not one-off campaigns—to compress the time from insight to execution.

- Use AI to synthesize calls, CRM notes, and web behavior into objections, proofs, and enablement updates.

- Treat positioning as adaptive: test variants by segment and learn from “second-best” messages.

- Maintain brand consistency by training models on guidelines and anchoring content in a reusable bank of proofs.

- Curate a pragmatic AI stack: pick tools by task, leverage custom GPTs/LLMs, and document what works.

- Make test-and-learn cultural: start small, secure executive sponsorship, and run weekly operating rhythms.