Skip to main content
for founders

Founder judgment in the AI era

AI can produce options. It cannot make the call. The founder's job is the call. Here is how to protect the muscle.

Updated Reviewed by Senwitt Editorial Team

How do founders keep their judgment sharp while leaning on AI?

Decide before you ask. Use AI for option generation, market research, and rough drafts; do the decision yourself first, then use AI to pressure-test it. The originating cognitive act of judgment under uncertainty is the founder's job; outsourcing it to AI quietly weakens the muscle the role most depends on.

Most founders in 2026 run on AI. Strategy memos run through a model. Customer interviews get summarised. Hiring shortlists get pre-screened. Investor updates get drafted in a window and edited in three. The productivity case is obvious and the tools aren't going away.

The narrower question is the one this post is for. What cognitive acts does a founder need to keep doing themselves, every working day, so that the AI integration doesn't hollow out the part of the job that defines the role — judgment under uncertainty, deciding before knowing?

Why this matters — the published evidence

The 2025 MIT Media Lab preprint by Kosmyna et al. (arXiv 2506.08872) is the cleanest recent empirical reference. It measured EEG, recall, and self-reported ownership across essay writers using brain-only, search-engine, and LLM-assisted conditions. The LLM group showed the weakest neural engagement during the originating task and the worst recall of their own arguments afterward. It is a preprint, the Stanković 2026 commentary raises methodological objections, and neither paper is the final word. But the direction matters for founders because the originating-act pattern in writing transfers to the originating-act pattern in deciding.

The 2024 MDPI Societies study (MDPI) extended the framing to AI in knowledge work and found an inverse correlation between AI usage frequency and self-reported critical-thinking engagement. The older cognitive-offloading literature (Risko & Gilbert, 2016, Trends in Cognitive Sciences) is the foundational cite for why this pattern exists at all — cognitive acts that get fewer reps tend to weaken.

The deliberate-practice literature (Ericsson, Krampe & Tesch-Römer, 1993) provides the constructive frame. Skill is maintained by daily, effortful, on-purpose engagement with the specific cognitive act. For a founder, the specific cognitive act is judgment.

The originating-decision act

Founder judgment lives in a narrow cognitive moment. You have incomplete information, time pressure, real stakes, and no one above you to defer to. You have to commit. The act of committing — under those conditions — is the load-bearing variable. Everything around it (research, options, drafts, sanity checks) is supporting work.

AI is excellent at supporting work. It can generate options, surface base rates, draft the memo, list the risks. None of that performs the commitment. But the way AI workflows tend to be used in practice, the supporting work expands until it crowds out the originating act. You ask AI for the analysis. You ask it for the recommendation. You accept the recommendation. The decision happened — to the world it looks like you made it — but the cognitive act of judgment didn't.

The pattern in the writing data (Kosmyna 2025 preprint) is that LLM-assisted writers had the lowest self-reported ownership of essays they produced. The transfer to decisions is the part founders should think about. If the originating act of judgment is outsourced often enough, the ownership goes with it.

A daily practice for shipping-heavy schedules

The routine below is built for a real founder's calendar — back-to-back, AI-saturated, with no time for a dedicated cognitive workout. None of it requires you to stop using AI. All of it requires about fifteen minutes a day, mostly inside decisions you would have been making anyway.

1. Decide before you ask. For every non-trivial decision in your day, write a one-sentence call before opening AI. Even if the call is wrong, you have done the originating act. AI then becomes the pressure test, not the originator.

2. Make one call a day with no AI in the loop. A hiring read, a feature trade-off, a customer prioritisation. Small enough that you can defend the decision without the tool, big enough to count as a real cognitive act. The point is the daily existence of the unmediated act.

3. Articulate the reasoning before reading the AI version. When you do ask AI for analysis, write your own version of the reasoning first, even just three bullets. Then compare. The comparison is the rep; reading the AI version cold is not.

4. Run a weekly judgment review. Once a week, review three decisions from the previous week. Which ones held up? Which ones were wrong? Which ones were AI-led versus founder-led? The review is what keeps the calibration honest over months.

5. Protect a no-AI thinking block. Twenty-five minutes a day, on the calendar, with no AI open. Use it for the hardest pending decision. The window does not need to produce a shippable artefact. It just needs to exist.

What this is not

A few honest disclaimers, because the founder category in particular is full of overclaim about AI and cognition.

This routine does not claim that AI use causes general cognitive decline or any clinical condition. The published evidence does not support that claim. The narrower claim — that the originating act of judgment, like any cognitive act, weakens when it gets fewer reps — is what the cognitive-offloading literature supports, and what this routine protects against.

It does not claim founders should use less AI. The productivity gains are real and there is no version of 2026 founder life where ignoring AI tools is competitive. The recommendation is calibration, not abstinence.

It is not a substitute for hiring well, building a team that gives honest feedback, or having advisors. The judgment muscle is one variable in founder performance; this routine protects that one variable, not all of them.

How Senwitt fits

Senwitt is a daily seven-minute Set across six Senwitt Skills — Writing, Math, Code, Memory, Reading, and Reasoning. For founders the Reasoning and Writing skills carry most of the load, but the mix is the point. The deliberate-practice frame Senwitt is built on comes directly from the Ericsson 1993 paper — daily, effortful, on-purpose engagement with the specific skill.

The for founders page lays out the case in more detail. The daily set page explains the design. For the longer argument on what the cognitive-offloading research actually supports, the research/cognitive-offloading page is the canonical reference.

A final framing worth carrying out of this post. Founder judgment is one of the few cognitive surfaces in business where the feedback loop is genuinely long. A bad early decision can take months or quarters to expose; a good one can take longer than that to look right. That length is precisely why the offloading risk matters more here than in disciplines with faster feedback. A software engineer who lets an AI write the wrong function gets a failing test within minutes. A founder who lets an AI talk them out of a non-obvious call finds out four product cycles later, when the cohort data finally comes in. The slow loop means the founder cannot rely on external correction to keep the originating judgment muscle trained. The training has to happen inside the daily practice itself. The cognitive-offloading literature is, in that sense, more directly load-bearing for the founder case than for almost any other knowledge-work role: the cost of the missed reps is real, the discovery window is long, and the only place to install the daily practice is on your own calendar, before the company is mature enough to make it standard.

From Senwitt · advertisement

The text above is editorial. What follows is a promotional message from Senwitt, the maker of this site. Senwitt is a brain-exercise app and is not a medical product. Read the full disclaimer in the footer.

Sources

  1. 1.Your Brain on ChatGPT: Accumulation of Cognitive Debt (preprint) arXiv, 2025.
  2. 2.AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking MDPI Societies, 2025.
  3. 3.The Role of Deliberate Practice in the Acquisition of Expert Performance Psychological Review 100(3):363–406, 1993.
  4. 4.Cognitive Offloading Trends in Cognitive Sciences 20(9):676–688 (DOI 10.1016/j.tics.2016.07.002), 2016.
Get the app

Take this argument with you. Daily practice in the app.

Download on the App StoreGet it on Google Play

Free download. Super Senwitt available in-app.

We use cookies to make the site work, measure aggregate usage, and (if you opt in) attribute organic app installs. You can accept all, reject all, or customize.

See our cookie policy and privacy policy.