In October 2026, near the end of an ordinary paid call through an expert network, a client asked how I stay current on the industry. My answer turned into the first request I’ve seen that the network’s compliance team had no process for.
I didn’t answer with a list of trade publications. I told the client about a weekly digest I built for myself. It’s an AI agent that reads a curated set of sources early every Monday, sorts the prior week into fixed lanes, writes a five-minute briefing with a source link on every item, and drops it in my inbox. It runs on my own setup and on the AI subscription I was already paying for. It added nothing to my monthly software bill.
The client wanted to see how it worked. The network set up a short paid written engagement, and I delivered a plain overview: what the digest does, the steps it runs, the design choices that keep it short, and the kinds of sources it reads. I also listed what a deeper build would involve.
Then the follow-up stalled.
For several business days, the question of how to take it further sat with the network’s compliance team. The account rep apologized more than once and was straight with me: they had never handled a request like this, and they were building the process as they went.
Even the part that was already finished got tangled. Over the following week, I was asked for the same written work four separate times, once in the original request and three more times through automated emails, and I submitted it twice. The work hadn’t changed. The system just didn’t seem to know where it was supposed to go.
Nothing about the request was hard
From my side, this was easy work. The system already exists. Pointing it at a different industry is mostly a matter of changing the source list and the lanes.
From the client’s side, it was a small ask. They wanted to know how one operator keeps up, and whether their team could do the same.
And the network wanted the work. Nobody in this story was trying to slow it down. Everyone wanted the same outcome, and the process had no slot for it.
The network sells what I know, not what I built
Expert networks are built around a clean idea. A client pays for an hour of someone’s knowledge. The expert talks, the call ends, and the transfer is over. Compliance is built to police that exchange: nothing confidential from a current or former employer, nothing material and non-public, no side deals.
A written answer fits that model. So does a survey.
A working system doesn’t. Once the client asked how they could run something like my digest themselves, the request stopped being about what I know and became about something I made. That raises questions an hourly-call process was never designed to answer. Who owns a build guide? Is it a consultation or a license? How do you price an hour when the deliverable is a method someone keeps using for years? Does handing it over start an ongoing relationship the network now has to watch?
I don’t know how they’ll answer those. I suspect they don’t yet either. That is the whole point.
The case for going slow
The strongest argument against my impatience is that compliance is doing exactly its job. The expert network industry went through insider-trading prosecutions a little over a decade ago, and the business rests on clients trusting that nothing improper passes through a call. A new kind of deliverable is precisely where a careful compliance team should slow down. If I ran that team, I would want to see the thing before I priced it, too.
I agree with all of that. My point is narrower. The delay isn’t a sign that anyone did something wrong. It’s a sign that a category is missing.
Experts are starting to show up with tools
Here is what I think is changing. Until recently, an operator showed up to an expert call with experience and a point of view. Now some of us show up with experience, a point of view, and software we built for our own practice.
I have no developer background. Since February 2026 I’ve built, with AI agents, a small company’s worth of tools, mostly for my own use. The digest is one of the smallest. It does a job I would otherwise cover with a paid monitoring service or with hours of my own reading every week.
I won’t be unusual for long. When a retired plant manager, a former buyer, or a one-time finance lead can build a research tool in an evening for close to nothing, the question “how do you stay current?” starts to have answers that are things, not habits. Clients will ask for the things.
Networks will need a box for that: a way to scope it, price it, check it for anything that should never leave the expert’s head, and hand it over cleanly. I expect the first network that builds that box well will win work the others end up turning away.
Where I’m drawing my own lines
Until that box exists, I keep a few rules for myself.
- I describe the system, never the client work it came out of. This post names no network, no client, and nothing said on the call.
- The full source watchlist and the working instructions stay mine. What I hand over explains how it works and why, at a level a team can act on.
- Nothing in the tool, or in anything I deliver about it, comes from a former employer. It reads public publications and company pages.
- Any version I build for a client I met through a network goes through that network.
Where this breaks
This is one request, at one network, in one industry. One slow ticket is an anecdote, not a trend, and I may be reading too much into a slow week.
It did end. A little over a week after I first sent the written work, I followed up once more, someone more senior stepped in, and the payment went through shortly after. What I still don’t know is whether the follow-on, the part that would have meant handing over the system itself, will ever get a process of its own.
It also leans on an easy case. A news digest reads public sources and nothing else. A tool built on a client’s own data, or on anything close to an expert’s former employer, is a much harder handoff, and a network would be right to move even slower with it.
There’s a bigger version of this that I keep coming back to. My digest reads public sources. The same machinery could read an expert’s own captured knowledge instead: twenty years of decisions, failures, and fixes, saved as they happen and turned into a recurring briefing. That’s a very different tool, and a much harder compliance question. It’s where I’m headed next.
If you’re an operator thinking about building something like this for your own practice, the intro call is free and fifteen minutes.