Manifesto
We don't need another tool for remembering people. We need one for showing up.
A note from the people building Spark about why this app exists, what we think the existing category gets wrong, and what we're actually trying to do.
Every personal CRM in existence is, more or less, the same app. A list of people. A box to type a note. Tags. Filters. A reminder system. Maybe a timeline. The good ones are well-designed spreadsheets dressed up as relationship tools. The bad ones are just spreadsheets.
They all share a quiet assumption: that the problem with your network is that you can't remember things about the people in it. So they give you a place to store facts. Pets. Spouses. Birthdays. That one conversation about the ski trip.
We don't think that's the actual problem.
The actual problem is that you already know things about the people in your life — quite a lot, actually — and you still don't reach out, because knowingisn't what's missing. Actingis. Nobody opens a relationship app at 4pm on a Tuesday and thinks "you know what would be good right now, some data entry." Nobody.
The gap
The problem isn't memory. It's the gap between memory and action.
You can remember that your client Sarah's daughter applied to art schools. You can remember it perfectly. The problem is that at the exact moment you could do something useful with that knowledge — text Sarah, ask how it's going, become the friend who remembered — you're in traffic, or in a meeting, or doomscrolling, or simply not thinking about Sarah. And by the time you are thinking about Sarah, the moment's gone.
The gap isn't knowledge. The gap is between what you know and the right moment to use it. A note-taking app can't close that gap. Only a system that watches the calendar, watches the cadence, watches your relationship graph, and reaches out to you when something is worth doing, can close it.
That system has to be patient. It has to know when not to bother you. It has to be willing to say "today is quiet, here's nothing" — and mean it. Because if it shouts every day, you'll stop listening, and the gap will widen.
What we're building toward
What an AI in a relationship app should actually do.
For most of the last two years, AI in productivity apps has meant one of two things: a chatbot bolted onto the side of the screen, or a button that "summarizes" something you didn't need summarized.
Neither of those things solves the gap. We think the AI in a relationship app should do three specific things, and nothing else.
First, it should file. When you tell it something about someone — by voice, by text, anywhere in the app — it should write the right note on the right contact, link the right relationships, and update the right dates. You should never be the person filling in fields about people you already know.
Second, it should curate. Every morning, it should look at everything it knows about your network and tell you the three to seven people worth thinking about today. Not 80. Not an inbox. Not a backlog of guilt. A short, ranked list of who matters right now, with a drafted opening for each.
Third, it should learn.When you ignore a suggestion or dismiss a draft, it should ask why, and remember. Most apps treat "no" as silence. We think it's the most honest signal the user gives.
File. Curate. Learn. That's the whole thing.
Beliefs
A few things we believe.
Low-volume beats high-volume.
A relationship app that ships 40 notifications a week is not a relationship app. It's a Slack channel for your guilt. Spark sends one push a day, and only when there's something worth pushing.
Each day is its own thing.
Yesterday's unacted-on suggestions don't roll forward and pile up. They expire. Tomorrow gets a fresh list, ranked by what actually matters tomorrow. Inaction is data, not debt.
Cadence is sacred.
A relationship has a natural rhythm. Some people are weekly. Some are quarterly. Some are once-a-year-on-their- birthday. We'd rather wait three months to surface the right person than nag you about a wrong one tomorrow.
Talk, don't fill.
If you have to choose between forms and conversation, pick conversation every time. The app should be something you talk to, not something you operate.
Privacy is positioning.
We don't scrape your inbox. We don't read your texts. We don't sell anything to anyone. The product is the subscription, not the data. If we ever change our minds about that, the company is dead — and we know it.
Every AI action is yours to undo.
The AI in Spark is autonomous, but never irreversible. Every note it writes, every contact it merges, every relationship it infers — undo button, 24 hours, no questions asked. Trust is earned by being correctable.
Who Spark is for
People whose careers run on relationships.
Spark isn't for the "personal life" market. It's not a journaling app. It's not a dating spreadsheet. It's not for people who want to "optimize their friendships."
It's for the realtor who can name 400 past clients but hasn't talked to most of them in two years. It's for the contractor whose next 30 jobs will come from referrals from his last 30 jobs. It's for the photographer who books her year from a sphere of 80 wedding planners and venues, and who can't afford to let any of those 80 cool off. It's for the financial advisor, the wealth manager, the consultant, the agency owner, the insurance broker — anyone whose income next year depends on the warmth of the people in their phone today.
These are professionals running networks. Not consumers keeping a contact list. Spark is the AI relationship manager for that work specifically — for the quiet, compounding, multi-year work of being known.
If any of that resonates — if you've ever opened a spreadsheet of people you mean to call and closed it again without calling anyone — try Spark for two weeks. We think it'll be the first app in this category that you actually keep using past day three.
— The Spark team