The market is full of clones of workplace tools. After trying every one of them I could find, I built a quieter alternative. Here is what was wrong, and what I changed.
A few years ago I forgot the name of my friend's dog.
This is not a serious problem, the way "I forgot my anniversary" is a serious problem. The dog is small and brown and has a name I am almost certain begins with M. My friend will not stop loving me over this. But the small failure stayed with me long after the conversation, because it was an instance of a category of failure I had been quietly accumulating: I was losing the small details about the people I cared about most.
The name of a coworker's mother, before her surgery. The book my closest friend mentioned in passing, the one I meant to read so we could talk about it. The week my brother started a new job. A friend's miscarriage, which I had been told about with great care and then promptly let slip out of working memory along with seventy unread Slack messages.
I am not a worse person than average. But the modern adult life — meetings, side projects, a phone full of attention-seizing apps — does not leave a lot of cognitive room for the small private details of other people's lives. The people you love stay loved. The texture of loving them slowly degrades.
So I went looking for a tool.
The personal CRM category exists. It has a name. There are venture-funded companies in it. I tried five of them, and none of them fit.
Dex was the polished default. Several thousand contacts, LinkedIn integration, tags like investor and customer, a "you haven't reached out to this person in 60 days" notification system. It is genuinely well-built for a specific user: the operator running a startup who needs to remember many people moderately well. I am not that user. I have approximately fifteen people I want to remember extremely well, and the surface area of Dex made my fifteen feel diluted in a database of strangers.
Clay was beautiful. The aesthetic — soft palettes, careful typography, the macOS app feeling like part of the OS — is genuinely lovely. The product's defining feature is that it pulls in your email and calendar and uses an LLM to write paragraph summaries of each person. I tried this. The summaries were technically accurate and emotionally cold, like reading a CRM's polite description of a friend. The act of not writing the sentence about my friend was, it turned out, the thing that prevented me from remembering. The LLM was doing my homework for me.
Folk CRM was a workplace tool with a softer aesthetic. Pipelines, deals, teams, email sequences. Useful for an agency; surreal for a friendship. I do not want to "sequence a follow-up" to my mom.
Monaru closed mid-2024. Tend sent me daily nag notifications until I deleted it. Garden had a literal plant metaphor where you "watered" your friendships and I felt insane every time I opened it.
I tried Notion. Notion is a database, and I wanted a notebook. I tried a paper journal. The paper journal was good but did not surface anyone on a Tuesday morning when I had three minutes between meetings to remember that my friend's dad had just been diagnosed.
What I wanted, it turned out, was something that did not exist.
Looking at the category as a whole, four mistakes kept appearing:
1. They were built around a network, not around a small circle. The defaults assume you have hundreds of contacts. The UI is optimized for the long tail. The pricing assumes you're paying with a corporate card. I have fifteen people. Most people have fifteen people. The networker is a real but small subset of the audience.
2. They borrowed vocabulary from sales. Words like "lead," "stage," "engagement," "pipeline," "follow-up cadence." Every time you open the app, the language nudges you into a frame where your friend is a unit of work. The frame is wrong. Friendship is not work.
3. They wanted to do the remembering for you. AI summarizers, auto-pulled email feeds, calendar inference. The hidden assumption is that the database having information about the person is the value. It isn't. The value is in you knowing the person, and that requires you to do the small effort of writing the sentence yourself. The minute the AI does it, you've outsourced the relationship.
4. They put your private journal in someone else's cloud. Every personal CRM I tried defaulted to "your data lives on our servers, behind our terms of service." For a corporate CRM this is fine — your sales pipeline is not that private. For a journal about my friend's grief, it is not fine. I do not want a startup's database admin to have technical access to my notes about my mom.
I spent eight weeks writing what became Weft.
The decisions were almost all negative-space decisions — what to leave out — and the product is shaped by them.
The product is designed around a small circle. The free tier supports seven people and Premium supports unlimited people. There is no contact import: you add people one at a time, and that friction is intentional. If your real goal is to enrich and manage hundreds of professional contacts, Weft is still the wrong tool; use Dex.
The vocabulary is journal vocabulary. Notes, threads, rhythm. There is no "stage" or "engagement score." Today keeps a short ranked list based on due threads, birthdays, and relationship rhythm, using language like "It's been a few weeks" rather than "OVERDUE." An optional generic daily reminder can run locally at the hour you choose, without streaks or guilt.
There is no cloud AI. No cloud LLM reads or writes notes, and there is no auto-summary or "write a follow-up message for you." The whole premise of the product is that you are the one doing the remembering.
The data starts in an on-device database. Weft uses a local SwiftData store. When Premium iCloud sync is enabled in Weft, synced records use the user's private CloudKit database; Apple encrypts those records in transit and at rest. Weft has no notes backend and cannot access private CloudKit records. There is no analytics SDK or third-party telemetry.
It costs $18.99 a year — or $39.99 once, forever, or free for up to seven people forever. The networker CRMs charge $120–$200/year because they're priced for sales reps with expense accounts. Most personal CRM users are paying out of pocket. The pricing should reflect that.
The honest version of this essay would also list the things I cannot do.
Weft is iOS-only. Not iOS + iPad + Mac + Web — just iOS 26, deeply native. If you live on Android, Weft is unavailable, and probably will be for years. I chose depth over reach.
There is no integration with LinkedIn, Gmail, calendar, contacts. The integrations are powerful, but the price is that your private notes get fused with your work data, and the texture changes. Weft is a clean slate; you bring what you want.
There is no team workspace, no sharing, no public profile. You cannot send a friend a link to "what Weft says about them" because the answer should be nothing — Weft is private, by design.
There are no streaks, no badges, no leaderboards, no celebration animations when you hit "Mark caught up." Friendship is not a workout plan.
If you've read this far and you're nodding along — if you also have fifteen people, and you also want a small quiet place to write a sentence about them on a Tuesday morning — Weft is what I made. It launches on iOS 26.
If you're a founder building a Rolodex of thousands of LinkedIn connections, you should buy Dex instead. If you want AI to summarize emails for you, you should buy Clay. If you're running outbound sales for an agency, you want Folk CRM. These are good products for the people they're for. They were just not for me.
If you decide to try Weft: free up to seven people, forever. No Weft account, no card, no email. The dog's name was probably Milo.
Try Weft. A quiet personal CRM for iOS 26. Free for up to 7 people. Get it on the App Store →