You are not bad at staying in touch. You just built a network before you built a system for remembering it.
Professional networks shrink faster than people think, quietly losing touch with hundreds of contacts every couple of years, according to Harvard Business Review research on the subject. Nobody decides to let a relationship go cold. It just happens, because the only system most of us have for tracking a network is memory, and memory does not scale past a few dozen people.
In the age of AI, that is a solved problem, not a personality flaw. So here is the exact system I built to fix it, using two ingredients: a structured database and an AI assistant that knows exactly what to do with what you tell it. No CRM software, no data entry job, just the tools already in your stack.
🎁 Reader perk: Get 3 months of Wispr Flow free, on me, using my partner link below.
Give Claude access to your Notion workspace through its connector settings, once, before the database even exists. The assistant is about to build the CRM for you, not the other way around.
Design: the database
Skip clicking through Notion setting up properties by hand. Describe what you want in plain language and have Claude build it directly: name, phone, email, LinkedIn, company, role, a relationship type you define (client, investor, press, vendor, personal), a status (new, active, follow-up needed, dormant), how you met them, and last interaction date. Leave room in the body of each entry for a running, dated interaction log, most recent first. That log is what makes it a CRM and not just a contact list.
Capture: the "add a contact" skill
This is the core of the system, a saved instruction set that runs every time you log someone. Pull whatever is available and do not require every field. Check for duplicates first, on phone number, then name plus company. A new contact gets a page, a "New" status, and a first log entry. An existing contact gets a new dated entry on top, never a deleted one. Confirm in one short line, no follow-up questions. Speed is the entire point.
Retrieve: the "look someone up" skill
A second, read-only skill for the other half of the job. Ask a plain question, "what do I know about Sarah," "who do I still need to follow up with from that conference," and get an answer pulled from the database and the log in natural language, not a raw property dump. If a name matches more than one contact, it should ask which one rather than guess.
Speak it in: Wispr Flow
The real unlock is not the database, it is how little effort it takes to feed it. A quick typed note, a forwarded screenshot of a contact card, or a voice memo dictated straight into Claude should all trigger the same skill. Wispr Flow is what makes the voice version frictionless, turning a rambling verbal note into clean text the second you stop talking, so updating the CRM never competes with whatever you are actually doing that day.
Test: real contacts
Add five or six real people. Look a few of them up. See where the auto-filled categories do not match how you actually think about your own network, and adjust the field options and instructions accordingly. This is the step that turns a generic template into something that actually fits how you work.
Automate (optional next step)
Once the manual version is running and you trust it, the natural add-on is pulling contacts automatically from sources you already use, email threads, call transcripts, instead of typing or forwarding them yourself. Get the manual workflow solid first. Automating a broken process just makes the mistakes faster.
That is the full system, connect to closed loop, one database, two skills, zero new apps to learn. The pattern here, a structured database plus a skill that knows your process, is the same blueprint behind every AI workflow worth building. Once you have built one, the rest are variations on the same idea.
If you want more of how I think about building in the age of AI, follow me on LinkedIn, Instagram, and TikTok so you do not miss what is coming next.
Jenny
P.S. If you want a weekly version of this specifically focused on AI, my other newsletter, The Women AI Weekly, is built exactly for that.