On Monday I asked an AI to find me people.
The tool is called Boardy, and it is an AI that makes introductions. You tell it who you want to meet, and it finds people who fit and introduces you once both sides say yes. I wrote it one careful email: the program I run, who it is for, and what I offer in return. Boardy answered in two minutes with a name.
By the end of the day it had asked me for something else twice. It wanted fifteen to twenty owners from my own network so it could write to them for me.
I said no both times. I would contact my own people myself.
Then it asked for two customer references, and I copied in two clients so they could decide for themselves. The next morning it hit me: I had put real people in front of a tool I had never actually checked.
On Tuesday I ran the check I give the operators in my course. It is five questions, asked one at a time.
What is the job? Introductions, nothing more.
What data would it touch? Today, mostly public information about my program. Later, maybe clients who could hire me, and maybe alumni from my business school. My answer was “it depends.” That turned out to be the most important thing I said. “It depends” is how a tool ends up holding more than you ever approved.
So the data counted as the most sensitive kind, and four rules followed this:
- Nobody goes in until they’ve said yes.
- No list, ever.
- No client’s business.
- No details from the alumni association’s records.
Then came the part I didn’t have time for. I was busy, so I asked Claude to read the policies.
The pages labeled Privacy and Terms on Boardy’s site were marketing pages. Friendly, short, and none of the four answers were on them. The real policies were one link deeper.
There, written in Boardy’s own words:
- A breach notice, in writing.
- Contact details shared only after “both parties have explicitly opted in.”
- “No training of AI models on proprietary business information without consent.” That line sits under the section for business customers.
The same policy also says:
- Conversations are used to “improve our services” and kept for three years.
- They are processed by “OpenAI, Anthropic, and other AI technology providers.”
- “The Boardy team has access to and monitors” what you say to it.
- Your email may be sent, in hashed form, to Meta to target ads.
I couldn’t find a full list of who receives the data, or a way to switch off the improvement use. Under my own rules, if I can’t find an answer, it counts as a no.
An introduction network is only as valuable as the people in it. The list request was not a trick. It was the business model asking politely. Every name you hand over grows the network, whether or not that person ever chose to join.
That is also why the friendly page sits on top. It isn’t lying. It is the version made for people who won’t click twice.
None of this is new. Long before any app, there was a rule about introductions: you don’t give out someone’s number without asking them first. Boardy’s double opt-in is that rule, written down. My no to the list was the same rule. I just hadn’t written it down.
Boardy passed with a limit, and the limit is one sentence:
Boardy hears what I’m looking for and who I am. It never gets a list, a client’s business, or an alum’s details. A person reaches Boardy only by saying yes first.
I still use it. The two clients I copied in had said yes first, so that part holds up. What changed is that the rule I followed on instinct on Monday is now written down where I can check it.
You are going to be asked to connect a new tool to your inbox, your contacts, or your clients this month. You can decide from the friendly page. Or you can spend twenty minutes on five questions and read one link deeper.
The check, free to install: https://github.com/WaziriG/skill-drops/tree/main/03-vet-a-tool
— Waz
Sources: Boardy, Privacy Policy (last updated June 26, 2026) and Terms and Conditions (last updated September 28, 2026), both read October 21, 2026. Policies change; check the current versions.