Context

Bob Safian talks with Axios CEO Jim VandeHei, co-author of the new book "Simplify," about why AI is making work life more complicated in the short term rather than less, and the three-part system (Confront, Delete, Amplify) VandeHei has run Axios on for the past two and a half years. The conversation matters to PMs and product leaders because it's a concrete case study in deliberately killing products and process, including Axios's own app, rather than layering AI on top of an already-cluttered operation.

The Big Idea

AI doesn't simplify work, it mass-produces more options, decisions, and noise, so the antidote isn't a smarter tool, it's a ruthless, repeatable system for confronting what you're doing out of habit, deleting what doesn't earn its place, and amplifying what's left.

VandeHei's evidence is his own company: Axios runs on this three-step system, and he cites data suggesting people waste roughly 25% of their work week on things that add no value, purely out of habit or inertia, a gap AI tends to widen rather than close since it lets anyone generate more emails, decks, and "well-researched" ideas than before.

Key Insights

AI is a complexity-creation machine in the short term

VandeHei calls himself an "AI lab rat" and admits he falls into the same rabbit holes he warns about. His core claim is that because AI lets anyone write well, research quickly, and generate polished-looking output, everyone downstream now receives more emails, more decks, and more "ideas," many of them not actually well thought out, which means more decisions land on everyone's desk, not fewer. He expects this to improve over time but is clear it is actively getting worse right now, precisely when people expect a productivity tool to be making things easier.

Confront, Delete, Amplify is a three-step ritual, not a one-time cleanup

VandeHei's framework runs in a fixed order. Confront means asking, out loud, why a given meeting, habit, or product exists at all, a question he says almost nobody asks by default ("do I need to be at that meeting?"). Delete is the concrete follow-up: once you've confronted a habit, decide whether to cut it or say no to it. Amplify only comes after the first two: once low-value obligations are actually removed, redirect that freed time and attention toward the work you're good at and actually want to do. He frames this as a repeatable system, not a single audit, because without a standing process, organizations default back to inertia.

Cognitive fatigue makes every extra decision cost more than it looks like

VandeHei cites brain-scan research showing that the more decisions a person makes, the worse they get at making decisions, a phenomenon called cognitive fatigue. He estimates the average person now receives around 120 texts and 150 emails a day, and argues that even a decision as small as choosing to ignore or delete a message still counts as a decision that draws down the same finite capacity. His practical response is automating away small recurring decisions entirely (he cites auto-ordering household essentials like toothpaste) specifically to reduce the total decision count, not to save money.

Killing the Axios app was resisted for years before the math won

Axios debated shutting down its own mobile app for years internally. VandeHei's tipping point was recognizing that only around 30,000 people used the app regularly, out of tens of millions of Axios readers overall, while maintaining it (app-store updates, ongoing engineering time) consumed real, ongoing resources. Most readers, he found, actually preferred getting Axios content through social media, a newsletter, or the website rather than a dedicated app. Killing it freed up time, money, and mental space that went into strengthening what Axios is actually differentiated at, which he frames as the general payoff of deletion: doing fewer things means getting measurably better at the things you keep.

A licensing deal with a platform only works if you don't depend on it

Axios signed a content-licensing deal with OpenAI, trading the right to sue over ChatGPT training on its content for direct payment, while Google search traffic to Axios simultaneously collapsed from roughly 30 to 40% of total traffic down to 5 to 10%. VandeHei explicitly contrasts this with BuzzFeed's earlier bet on Facebook distribution, which he says "hasn't ended well," and states his own rule: never build your business model on the generosity of a third party with no vested stake in your survival. His deal works, in his framing, specifically because losing that OpenAI revenue tomorrow "doesn't make a damn bit of difference" to Axios's overall viability, since the core business was never dependent on it.

Local news becomes viable once AI removes non-journalism costs

VandeHei applies his own Confront/Delete/Amplify framework live to Axios's local news expansion: he confronts the assumption that local news is unsustainable (print costs, distribution, large newsroom overhead), identifies that AI can delete most of the non-journalism costs (content distribution across platforms, copy-editing support), and argues that removing those costs is what lets Axios amplify actual journalism into news-starved local markets. His stated goal is to hire literally hundreds more journalists as a result, arguing that AI is removing the surrounding cost structure that made local news unprofitable, not replacing the journalists themselves.

Mental Models & Frameworks

Confront, Delete, Amplify

A three-step, always-in-that-order system for cutting anything (a meeting, a habit, a product line, a personal social obligation) that no longer earns its place.

• Confront: explicitly ask why something exists, out loud, since almost nobody does this by default and organizations otherwise run on pure inertia.
• Delete: once confronted, actually cut it or decline it, rather than just noting that it seems unnecessary.
• Amplify: redirect the time and attention freed up by deletion into the work you're actually good at or that matters most, which only becomes possible once the first two steps have genuinely happened.

VandeHei uses it at every scale, from a single weekly meeting up through cutting an entire mobile app and reshaping a business line like local news, and argues the order matters: skipping straight to amplifying without first confronting and deleting just adds more activity on top of an already-cluttered baseline.

The three-things list as an alignment check

VandeHei has employees take ten minutes to write down, in order, the three things they believe they need to do to excel in their role, and separately the things they think are dumb or unnecessary that the company or they themselves do. He calls this "magical" for two reasons: comparing the employee's own list against what their manager would have written surfaces misalignment about priorities directly, and the "dumb things" list is usually accurate, since people who already suspect a process is wasteful are often right, and naming it explicitly opens the door to actually stopping it.

Practical Application

Run a "let's kill sh*t" meeting with your leadership team

VandeHei convened his leadership team specifically to list every meeting, process, and product they believed the company did that was unnecessary, then spent one to two hours cutting items from that list. Structure this as its own dedicated meeting, not a side agenda item on something else, and treat "we've always done it this way" as disqualifying rather than a reason to keep something.

Have every employee write a two-list exercise

Ask each person on your team to spend ten minutes writing, in priority order, the three things they believe they need to do to excel at their job, and separately the things they personally think are wasteful or better handled by someone else or not at all. Compare the first list against what you as their manager would have written to catch misalignment, and take the second list seriously rather than treating it as venting.

Automate small recurring decisions to reduce total decision load

Identify the small, repeated purchasing or logistics decisions in your life or team's workflow (VandeHei's example is auto-ordering household staples) and remove them entirely rather than just doing them faster. The goal specifically is lowering the total count of decisions you make in a day, since cognitive fatigue research shows decision quality degrades with volume regardless of how trivial any single decision seems.

Before adopting a new AI workflow, ask what it should let you delete

When adding an AI tool or capability to a process (as VandeHei did with Axios's local news cost structure), explicitly identify which existing cost, step, or team the new capability is meant to remove, not just what it adds. If a new AI tool only adds output (more drafts, more research, more options) without a corresponding deletion somewhere, it is adding to the complexity problem VandeHei describes rather than solving it.

Bottom Line

Jim VandeHei's central claim is that AI's near-term effect on work is more noise and more decisions, not less, so the real leverage isn't a smarter tool but a disciplined, repeatable habit of confronting what you do out of inertia, deleting it, and only then amplifying what's left, a system he's used to kill his own company's app and reshape its local news strategy.