Pick a company ethos on AI before the tools pick one for you
Drew Jonsen · Founder, Jonsen LLC July 27, 2026 4 min read
What work will your company never hand to an AI tool, and who has the authority to change that answer? The first half is usually easy to say out loud. The second half is where it goes quiet. Until both are written down, the position gets set for you, one small choice at a time, by whichever tool a well-meaning team member turned on last week.
A company AI ethos is a short, plain-language document that says where you stand. It's the internal reference every future AI decision gets measured against. It's written for your own people, so it doesn't need to read like a compliance policy or a press release.
What an AI ethos covers
A useful ethos is narrow enough to enforce and broad enough to survive the next model release. Four areas cover most of what a leadership team has to decide.
1. What gets automated and what stays human
Name the categories. Drafting, summarization, and internal search are common candidates for automation with a review step. Final legal judgment, hiring decisions, and client-facing communication above some threshold usually stay human. The line will move as the tools improve. Draw where it sits today, so a project team knows which side of it they're on before they start.
2. Transparency with clients and staff
Decide two things. Is your team expected to disclose AI use in client work? And can your clients ask for a description of the AI systems that touch their data? Leaving either one ambiguous is a trust problem waiting to happen. A conservative answer you have written down will serve you better than a generous one everyone assumes.
3. Data red lines
List the categories of information that must not leave your governed environment: privileged material, regulated health or financial data, HR files, unreleased strategic work. Pair each one with the handling that is allowed. This is the section your team reaches for the week a new consumer AI tool shows up with a free tier that's hard to say no to.
4. Who decides
Name the person or the committee that approves new AI tools, new data sources, and exceptions, and name the turnaround they owe an answer in. A body that meets monthly is how exceptions turn into shadow tools. How quickly you can decide is part of the ethos.
How to write one
The best AI ethos we've seen ran to a single page. It read like an operating document, something a team member could get through in three minutes and act on the same day.
Draft it with leadership, not for leadership. The people who will enforce it should be the ones writing it.
Use plain language. If a paragraph needs a glossary, rewrite the paragraph.
Give three concrete examples for each principle, because abstract rules get misapplied.
Set a review cadence. Quarterly works for most companies, since this document ages faster than a values statement does.
If leadership doesn't write down a position on AI, your vendors will write one for you.
How the ethos guides tool decisions
Once the ethos exists, tool selection gets faster and calmer. A new drafting assistant shows up, and the question is whether it fits your current position on automation, transparency, and data. If it does, it moves to pilot. If it doesn't, you decline it in an afternoon instead of a quarter.
The same document sets terms for the tools you already run. A general-purpose chatbot in an internal app gets clear boundaries around which sources it can read, which conversations it can retain, and which questions it should escalate. An automation platform picks up guardrails. And the consumer AI tool somebody has been quietly using lands on the exceptions list, where somebody has to make a call on it.
The version we recommend starting from
For most mid-size businesses, a first draft takes one working session and about a month of iteration. Start with the four sections above. Add examples pulled from your own last six months, since those are the ones people will recognize. Circulate it to the people who will use it every day, and not only to the people who sign off on it. Then publish it internally and refer to it when the next AI decision comes up.
The document belongs to leadership, with one named executive accountable for it and the approval body from section four keeping it current. That group revisits it quarterly, and out of cycle whenever a new tool or a new data source pushes on one of the lines it draws.
Key takeaways
Without a written ethos, your AI defaults get set by whichever vendor your team enabled last.
A good ethos covers automation scope, human-in-the-loop rules, transparency, and data red lines.
Keep it to one page in plain language, so the people who have to enforce it can actually use it.
Use the ethos as the acceptance test for every AI tool decision, big or small.
Ready for a calmer conversation about technology?
Drew replies personally within three business days.
Written by
DJ
Drew Jonsen
Founder, Jonsen LLC
Drew leads Jonsen LLC, a Denver technology practice guiding law firms and growing businesses through AI, cybersecurity, and systems that compound over time.
