The pilots stall. The dashboards multiply. Governance says no, so the business routes around it. I've spent my career fixing exactly this — building the data, AI, and governance foundations that let enterprises move fast with confidence, and killing everything that doesn't earn its place.
I'm Jonathan Salmon — Director of Enterprise Intelligence at Centene, leading a 15-person team in one of healthcare's most demanding data environments.
Different companies, same three failures. If any of these sound familiar, we should talk.
Hundreds of reports nobody opens, each one taxing the platform and the people who maintain it. Waste compounds.
AI demos that never reach production because nobody owns the path from experiment to product. The budget line is on borrowed time.
Controls designed to say no instead of enabling speed. So the business routes around them, and the real risk goes underground. Good governance is automated and invisible: guardrails, not gates.
What I've actually owned and delivered — the kind of scope a VP remit is built on.
Lead a 15-person enterprise intelligence organization inside a Fortune-scale healthcare company — regulated data, real stakes, no room for vanity metrics. Two years at director level and counting.
Own the platforms and the reporting estate. My rule is simple: every report and dashboard must earn its existence — or it goes. High-utility core products, built with best practices, replace sprawl.
Known for translating strategy into executable plans and empowering strong engineers rather than hovering over them. Talent development and mentoring are the multiplier on everything else.
Geodynix — U.S. market-intelligence product, live in production. Profit Sentinel — B2B SaaS for profit-leak intelligence. I don't just govern data platforms; I ship product on them.
The beliefs I actually manage by — not poster slogans.
Hire well, set direction, clear obstacles, get out of the way. Control is the enemy of scale.
Say the thing — to your team, your peers, and up. Ambiguity is where careers and projects quietly die.
Reports, dashboards, meetings, metrics: prove value or kill them. Waste compounds; so does focus.
Your team's growth is your output. Mentoring isn't charity work — it's how you build an organization that outlasts you.
Ask earlier. Advocate louder. Manage up on purpose. Power dynamics are real; pretending they aren't is not a strategy.
Most value concentrates at the edges; most effort bleeds into the middle. Find the vital few inputs, starve the trivial many, and treat waste as the enemy — because it is.
A straight path from reading to conversation.
Read the AI Signal — five stories every morning, each with my take. You'll know within a week whether I think the way you need.
Email me directly. I read everything myself, and I make time for good conversations.
Fractional, advisory, or board conversations. A straight conversation about your problem.
Five stories a day from a short list of sources I actually read — each with my one-line take. This is how I stay in the conversation.
Updated Friday, November 13, 2026
My take: Nvidia told Rep. Bill Foster privately that geotracking advanced AI chips to fight smuggling to China is technically possible, even as its industry group called the Chip Security Act mandates potentially infeasible in public. That gap between private capability and public position is exactly why governance cannot run on vendor assurances alone: require the mechanism, verify the location, and put it in writing.
My take: Gilbert + Tobin reached 87% active use of ChatGPT Enterprise, more than twice its normal tool adoption, by moving from safe experimentation to personal productivity to defined agentic workflows where people still approve the work. CEO Sam Nickless modeled use himself and set no usage mandate: leaders go first, governance enables rather than blocks, and the payoff is time returned to judgment, not fewer people.
My take: DeepMind positioned Gemini 4 Hydrogen as the fast execution engine alongside Carbon and Argon for judgment, priced at $1.50 per million input tokens, with a new Orchestrator that switches models mid task for near Carbon success at 24% less quota and 19% faster completion. That is the right enterprise pattern: do not spend frontier money on every step, route work to the cheapest model that can do it well.
My take: AWS shipped an official local emulator for its services, written in Rust by a swarm of agents that made 29,684 commits in three weeks, so teams can test without provisioning paid API keys. The emulator matters, but the build story matters more: agent volume is now real, and the leadership job shifts to review, verification, and knowing what should never have been built.
My take: incident.io shared rare real numbers: about $1.6 million in AI spend this year, now roughly $1,600 per engineer per month after Opus 5.5 cut costs from about $4,500, with nearly 80% of production PRs including agent generated code. The median engineer shipped only about 10% more PRs while tenured engineers shipped about 70% more, which is the 80/20 lesson in plain sight: AI amplifies people who already know the codebase, it does not replace that knowledge.
Founder work I run outside my day job — where I stay close to the craft of shipping.
U.S. market-intelligence product unifying demographic, economic, housing, and business data — sold as a Snowflake data product and a self-serve insights app.
B2B SaaS: continuous profit-leak intelligence for mid-market trade contractors, auditing the seams between their operational and financial systems.
I'm writing on AI in the enterprise, data leadership, and organizational power — publishing on Medium as part of a deliberate content practice. The AI Signal above is the daily version; the essays are the deep version.
Latest: AI Made Building Free. It Didn't Make Deciding Free. — why review, not output, is the scarce skill now.
I also keep a public reading list — the books that shaped how I think.
I make time for good conversations — with data and AI leaders, founders who want an advisor, or peers working on the same hard problems. I read everything myself.
jonathan.salmon@gmail.comRésumé available on request.
Jonathan Salmon