I Don't Use AI to Go Faster.
I Use It to Think Differently.

These two tools together have essentially replaced the research phase of every major PMM deliverable. NotebookLM is where I dump PMM content — analyst reports, Gong call transcripts, customer interview recordings, competitive battle cards, win/loss data — and query across all of it without hallucinating outside those sources.

In the cloud management and consulting space specifically, where the competitive set shifts constantly — MSPs repositioning around FinOps, hyperscaler-adjacent players expanding into professional services, niche tools getting absorbed into platform plays — manually tracking that movement used to mean a rotating stack of browser tabs, saved Google Alerts, and a CI doc that was stale before it was finished. Now I load a quarter's worth of competitor content, earnings call summaries, G2 review exports, and partner program announcements into a notebook and interrogate it like a structured dataset. The signal-to-noise ratio is completely different.

Perplexity handles real-time external research — competitive landscape shifts, analyst commentary, recent product announcements — with citations I can actually verify. Together they compress what used to be a two-day research phase into a three-hour synthesis session.

Long-form positioning work is where Claude's extended context window becomes a genuine structural advantage. I can feed in a full positioning brief, a set of customer interview quotes, competitive differentiation notes, and a persona profile — and work iteratively on messaging architecture without losing thread across sessions (Claude Projects preserves context). The quality of the thinking-partner conversation here is different from a general chatbot; it holds nuance, pushes back on weak logic, and produces copy that needs editing rather than rewriting.

This combination is the operational CI backbone. Klue's Compete Agent ingests Gong call recordings and CRM data to auto-update battlecards when competitors get mentioned in live deals. The feedback loop from field → deal intelligence → PMM → updated asset used to take weeks. Now it's days, sometimes hours. For a cloud SaaS company with a fast-moving competitive landscape, that latency reduction is a genuine GTM advantage.

Product launches require a volume of assets — campaign copy variants, social posts, partner co-marketing templates, event materials, internal enablement docs — that used to require either a large team or painful tradeoffs. I've been using Jasper trained on brand voice, which matters more than it sounds. The difference between generic AI copy and copy that actually sounds like you is almost entirely in the training layer. Once Jasper has internalized tone, vocabulary, and the specific way a brand talks to its audience, it stops producing output that needs to be rewritten from scratch and starts producing output that needs to be edited. That's a meaningful shift in the economics of content production. Canva AI handles the visual layer. Neither produces finished work without judgment applied, but together they mean a PMM can ship a launch asset kit that would have previously required a dedicated content team.

I'll be honest — this one I'm still experimenting with, and it's new enough to my stack that I won't overclaim on it. But the underlying problem it's solving is real enough to flag. Sixty percent of B2B buyers now use AI tools during vendor evaluation — ChatGPT, Perplexity, Claude. If your product doesn't appear in those AI-generated responses, you're invisible to a growing share of your ICP before they ever reach your website. Profound tracks where your brand appears (and where it doesn't) across AI engines, which is supposed to directly inform content strategy for AEO (Answer Engine Optimization). Most marketing teams don't have this in their stack yet. I'm testing whether the insight it surfaces is actually actionable at the pace a lean PMM team can move. Early read: the visibility data is interesting; the question is what you do with it.