NOTE: Demo visuals include blurred data or synthetic placeholders to protect customer privacy.

Product marketing has more questions than calendar slots for analyst calls. A skill that generates the dashboard in the moment, in the context of a specific prompt, beats a static report that ages out the week after it ships. Product Research Nexus is built around that bet.

Bringing new products to market in Abnormal's Act 3 means moving fast on pricing, packaging, naming, and competitive positioning. The traditional inputs (analyst conversations, vendor reports, customer surveys) remain valuable, but they don't fit the cycle time. Ryan saw the gap firsthand when product marketing and R&D both conducted competitor pricing research in the identity space and produced different numbers.

1. Manual primary research takes weeks. Analyst calls, vendor reports, and customer surveys are the standard playbook, but it's slow.

2. Parallel research surfaces conflicts. When PMM and R&D each ran identity competitor pricing, the results diverged, forcing a third pass to reconcile.

3. Static dashboards go stale. Building a fresh dashboard for every new question wasn't realistic, and reusing yesterday's view risks missing what changed this week.

Product Research Nexus is a Claude skill, not a hosted dashboard. Invoked as /product research nexus, it produces a new view each time, scoped to the prompt. Ryan trained the skill to score open internet searches higher than internal polls through Glean for competitor research, because, as he put it, "we can never be fully sure what we have internally is as accurate as what it can find and correlate in the open internet." The skill draws on Abnormal's AI-generated master messaging documents (MMDs), product context from Glean, customer signals from Gong via Glean, and live web searches against analyst sources and vendor coverage.

Capabilities Ryan demoed in his slot:

• Generates 12-month competitor news dashboards with an executive summary and a recency-weighted ranking of competitor pressure

• Builds a competitive landmines view that compares what competitors claim against what they actually shipped

• Auto-generates a magic quadrant score against criteria from the relevant product MMD

• Runs TAM analyses with funding, capability, and differentiation breakdowns, with inline source citations

• Produces pricing and packaging recommendations benchmarked against current Abnormal pricing

A small design pivot by teammate Mickey turned this from a static app that required an API key for every prompt into a just-in-time skill. That removed the maintenance overhead and made the dashboard disposable, which is the point.

The skill compresses the loop between question and answer. When the new products PMM team gets a competitive question in a meeting, they can run /product research nexus against the relevant MMD and have a sourced view inside the same session, not next week's deck.

For product marketing, the value is speed and reuse. The team can answer pricing, packaging, and competitive landmine questions in the meeting where they came up. For product and R&D leadership, the value is alignment.

Both groups can run the same skill against the same MMD, reducing the time spent reconciling each other's research.

• Cuts the "wait for the deck" cycle on competitive and pricing questions to minutes

• Surfaces messaging update recommendations alongside the competitor news that triggered them

• Cites sources inline when the prompt asks, so outputs are reviewable rather than trusted blind

• Reusable across the PMM team rather than locked to one builder

Ryan, Jason Yu in Product Ops, and Tanisha on the product team are now consolidating their parallel research and feature-ranking efforts into a single shared skill, so the org runs one good version instead of three adjacent ones.

The dashboard, in its earlier form, was a static artifact built by one person and maintained by another. Product Research Nexus inverts that. Ryan can ask a different question against the same data and get a totally different dashboard for a different audience, generated in the moment. A peer’s reaction was to take the pattern back to his own work.

The room's reception showed how the building style is changing at Abnormal. Mickey reshaped the asset from an app into a skill, which is what made it tractable in the first place. Maggie Kishibe surfaced that Jason Yu and Tanisha were already building adjacent versions and pushed for consolidation, so three independent threads were converging on one tool before the demo even closed. That is the operating mode Ryan was after when he said the constraint is no longer capability; it is how many ideas the team can carry.