OIObsidian Insights
Intelligence Roundup · For solo entrepreneurs

Intelligence Roundup · Solo edition

Turn AI into finished work.

Three new tools dropped overnight. Your client deliverable is still due Friday. Here's what actually deserves your attention.

In this issue

If you run a business on your own, every AI launch is a small tug on your attention. Testing tools feels like progress. But it doesn't deliver anything you can bill for.

Keeping up has quietly become a second job: dozens of channels, hours of video, and most of it isn't written for someone who is also the salesperson, the delivery team and the accountant. So I had my system do the watching. Out of everything it read, three ideas are worth a solo entrepreneur's time, and they fit together better than you'd expect.

Companion episode · 5:03 · invitation onlySolo Entrepreneur | Turn AI Into Finished WorkLena and Trevor talk through getting paid for outcomes, giving your AI your own context, and protecting the hours that earn you money.
Members: listen now Request an invitation

Signal one

Get paid for the outcome, not the effort

Outcome-based pricing came up again and again in what we read. There are real examples of it. Intercom prices its Fin support agent per outcome, and Zendesk charges for what it calls verified automated resolutions. In both cases the bill is tied to a support result rather than just the amount of AI used.

Read the definition, though. The provider decides what counts as an outcome, other fees or commitments can still apply, and none of it is a guarantee of revenue or profit. One recent discussion named the deeper catch: a system can hit the number you measure and still miss what the customer actually wanted. So define the result and the check before you agree on the price.

For a one-person business, that's a useful mirror. What result is your customer actually paying for? Not the hours, and not the tools. If you can't name it in one sentence, it's hard to delegate any part of it, to an AI or to a contractor.

A hypothetical to make it concrete. A freelance designer isn't really paid for “making slides.” They're paid for a pitch deck the client can present on Thursday without edits. Write that sentence down, and it becomes obvious which parts an AI could draft, like a first outline built from the client's own notes, and which parts stay with the designer: the story, the visual judgment, and the final call.

Signal two

Give it your context, not more words

One creator shared a practical fix for AI writing that sounds generic. Most people try to fix it with longer prompts. Instead, she keeps one folder with her own published posts, her content strategy, her brand guide and her goals for the year, and points the assistant at that folder. The assistant has examples of her real work to draw from, so the draft comes back closer to her voice.

A founder offered a version of this that takes almost no writing at all. When his team kept asking him the same questions, he started recording himself doing the task while talking through it, then had AI turn the recording into a written procedure. Same idea: your own work, captured once, becomes the reference material.

Two cautions keep this honest. Better reference material makes a draft more relevant, but it still needs checking, especially when a factual claim matters. And the folder isn't a security boundary. Permissions control what an assistant can reach; a person still reviews the draft before it goes out.

From my own work

At Obsidian Insights, I worked on the same problem from the other direction, with an orchestration skill and a refined approval ladder, so approved work carries its own limits instead of stopping to ask again. In my day-to-day use, I saw fewer stalls and fewer repeated approval requests.

That's my observation from working in the system, not a measured benchmark. Some stalls continued, and the full release is still unfinished. But the lesson matches hers: when the context and the limits travel with the work, the work moves.

Three steps in order: name what you're paid for, point AI at your own work, then protect the hours that earn money.
The three ideas, in the order they help. Each one makes the next easier.

Signal three

Protect the hours that earn you money

One entrepreneur coach put it bluntly: anything that makes you money, do it first thing in the morning. Writing a proposal, solving your biggest bottleneck: morning. Replying to email: afternoon, or, in his words, never.

A guest on a productivity podcast added a simple trick. Set a timer. When we can see time ticking by, we tend to focus and work faster. Parkinson's law came up too: work tends to expand to fit the time you give it. It's an idea to test, not a law of nature. Give a proposal a focused twenty-five minutes and see what happens.

For a solo business, that first protected block should go to one of three things: the deliverable a client is waiting on, the proposal that could become revenue, or the one bottleneck slowing everything else down. Admin can wait until the afternoon.

Just make sure it's the real bottleneck. One founder described being ready to redesign his whole program. Then he checked: he'd lost three customers out of three hundred that month. The number said the redesign could wait. Ten minutes of checking saved weeks of the wrong work.

The decisionWhat AI may doWhat stays with you
The draftAssemble it from your own folderEvery claim traces to something you wrote
Price and scopeQuote only what you've publishedAny new number or commitment
Dates and promisesNothingAll of them
SendingDraft and saveYou hit send

For your role

Make it yours

Protect the hours that earn you money, and give your AI your own context instead of longer prompts.

One thing to try this week

Try this: two small experiments

  1. Ten minutes. Choose one recurring task. Write down the customer outcome, the inputs it's allowed to use, how you'll check that it's good, and the point where you approve it.
  2. One twenty-five minute block. Protect it for a deliverable. Silence nonessential interruptions, keep your urgent contacts, and note at the end whether the deliverable got done.

These are proposals, not promised productivity gains. Run them once and see what you learn.

And if you're feeling behind on AI right now, you're not alone. One innovation advisor made the point that feeling behind can feel terrifying, but you can leapfrog really quickly. It's mostly a mindset problem. You don't need every new tool. You need one finished deliverable, done well, this week.

Invitation-only podcast pilot

Stop chasing feeds. Get the signal for your role in five minutes.

Staying current on AI now means a few dozen channels and hours of video every week, and most of it isn't meant for your job. So I built a private podcast that does the watching for you. Each week we review a myriad of AI channels and feeds, keep only what matters for your role, check each claim against its source, and turn it into a roughly five-minute episode: one for solo entrepreneurs, one for tech leads, one for client-facing teams.

This edition is the written companion to the solo episode. The audio is invitation-only while the pilot runs.

What's the one task you'd hand off first, if you could name its outcome in a single sentence?Tell me in the comments on LinkedIn.

That's this edition of the intelligence roundup. Go build something your customers will thank you for.

—WWilliam R. McBride, Jr.

Sources and notes

The other editions

Editorial note: This edition is not sponsored. Speakers and channels are credited under Sources and notes rather than in the text. Vendor pricing terms are the vendors' own and can change; read their definitions before relying on them. Statements about my own systems describe observed behavior in my working environment, not quantified benchmarks. The companion episode uses AI-generated voices and is available to invited members.