AI Product Engineer · Fractional Product & AI Leadership
Joel Mercer
Product strategy and working software, from the same person.
I'm Joel Mercer, an AI product engineer and ex-VP of Product. I take AI-native products from zero to production where the decisions are high-stakes and have to stand up to scrutiny: government, defence, national security, and regulated enterprise. Fractional leadership, advisory, and hands-on build.
Currently taking on a small number of fractional, advisory, and build engagements.
Track record
- Led product for Logically's AI intelligence platform from seed to Series A, scaling it across government and enterprise.GovTech ↗︎
- Named #2 on Fast Company's Most Innovative Companies in AI, 2021.Fast Company ↗︎
- Helped create the narrative-intelligence category, with a top-right placement in Gartner's Emerging Market Quadrant.Gartner ↗︎
- Cut analyst triage time by 60% with agentic, AI-assisted workflows.
- Finalist for the UK's Most Innovative Cyber SME 2021, run by DCMS and Infosecurity Europe.Infosecurity Europe ↗︎
- Quoted in The Hill on using AI to counter misinformation, as Logically's Head of Product.The Hill ↗︎
- Spoke on using AI to combat the disinformation crisis.Watch ↗︎
- Built VIXR, LLM-driven volatility intelligence for the S&P 500.vixr.ai ↗︎
What I do
Knowing what to build
Plenty of people can now build a tool. Far fewer can tell whether it should exist, how it should exist, and whether it survives contact with your regulators, your data, and your business. That is the scarce skill, and the one I lead with.
Strategy that ships, not slideware
I own the product direction and the AI architecture, then build the thing. The person setting the roadmap is the one who can write the pipeline, so nothing gets lost between the deck and the code.
Built for high-stakes decisions
Messy, conflicting, adversarial signals turned into evidence you can act on, with provenance, evaluation, and governance built in rather than bolted on. Answers that survive scrutiny.
Zero to a market
From customer discovery and rapid prototypes to paying customers, secure production, and a defensible category. I've done 0 to 1, and 1 to a market.
Where do you stand?
Is your AI ready to run where the decisions are high-stakes and have to survive scrutiny?
Take the 4-minute AI Production-Readiness Scorecard →Selected work
All projects →- AlbionIndependent project · 2026
A deterministic, LLM-free engine for detecting coordinated influence in social-media data.
- ArionIndependent project · 2026
A knowledge-graph platform that compresses weeks of foreign-influence (FIMI) analysis into hours.
- VIXR2025–present
Volatility intelligence for the S&P 500.
Writing
All posts →Knowing it should exist is the job
AI made building tools cheap. It didn't touch the part that was always hard: knowing what should exist, and whether it holds up when it matters.
I caught a coordinated influence campaign without spending a single token
A tech experiment in detecting coordinated social media activity through structural analysis, not content analysis.
Let's talk.
Tell me the decision you're trying to improve, or the product you're trying to ship. If I'm not the right fit, I'll say so.
Currently taking on a small number of fractional, advisory, and build engagements.