Semantic Search for Your Relationship Notes

Semantic search finds the right person or note by meaning, not by the exact words you typed. Ask “who do I know in climate?” and it surfaces the contact whose note says “joined a renewables fund” — even though you never wrote the word “climate.” That’s the difference between searching your relationship notes and only searching the strings inside them.

Keyword search is fine when you remember the precise phrase you used. The problem is that you usually don’t. You remember the gist — a field, a problem, a connection — not the literal words from a note you wrote nine months ago.

Keyword search vs semantic search

Keyword search matches characters. If your note says “supply chain” and you search “logistics,” you get nothing, even though they’re obviously related. Semantic search matches meaning, so a query and a note can connect even when no words overlap.

Your query	What keyword search needs	What semantic search does
”Who works in climate?”	The literal word “climate” in a note	Surfaces “renewables fund,” “carbon,” “grid storage"
"Someone who knows hiring”	The word “hiring”	Finds “scaled a team,” “head of talent,” “recruiting"
"People into endurance sports”	The exact phrase	Matches “ran a marathon,” “training for a triathlon"
"Who’s raising money?"	"raising” spelled that way	Finds “starting a seed round,” “fundraising in Q3"
"Anyone in supply chain?"	"supply chain” verbatim	Matches “logistics,” “procurement,” “fulfilment”

The right-hand column is how recall actually works in your head. You think “who did I meet who knows supply chain?” and you don’t care whether the note says logistics or procurement — you want the person.

Why this matters for relationships specifically

Relationship notes are written in the moment, in shorthand, in whatever words came to mind that day. You’ll describe the same idea ten different ways across a hundred people. Keyword search punishes that inconsistency. Semantic search forgives it.

Coffee with Aanya. Left consulting to run growth at a logistics startup, says fulfilment is their bottleneck. Training for her first ultramarathon. Asked if I know anyone who’s scaled a warehouse ops team.

Six months later you’re trying to remember “who was that person in supply chain who needed an ops hire?” You’d never recall that the note actually said “logistics” and “fulfilment.” Semantic search bridges the gap between how you remember and how you once wrote.

Good queries semantic search unlocks

Once search works by meaning, you start asking your notes questions you’d otherwise only ask a colleague:

• “Who did I meet who’s hiring engineers right now?”
• “Which contacts mentioned moving cities recently?”
• “Who’s connected to the fintech world?”
• “Anyone I owe an introduction to?”
• “Who’s going through something hard at the moment?”

These are recall questions, not lookups. They’re how you’d prep for a week of meetings if you had a perfect memory of everyone you know.

It has to stay grounded and private

This is where semantic search for relationship notes diverges sharply from web search. The whole value is that it searches your own notes — the things you personally observed and chose to record. It should not reach out to the web, public profiles, or third-party data to “improve” an answer.

Intriq applies this principle: you ask about the people you know in plain language, and its recall runs over the notes you saved — answering the question and citing the note it came from, instead of making you remember the exact word you typed nine months ago. Ask about something you never wrote down and it tells you it’s not in your notes rather than inventing a detail to fill the gap. The notes themselves stay local-first on your iPhone with encrypted on-device snapshots, and nothing about a person gets enriched or scraped from outside. The goal is to make your own memory more findable, not to import a stranger’s data into it.

That grounding is what keeps the feature trustworthy. Search by meaning is powerful, and power over real people’s information is exactly where you want the tool to stay honest about its sources and refuse to guess.

Semantic search vs the relationship graph

Semantic search answers “find me the person or note about X.” A relationship graph answers “how are these people connected to each other and to me.” They complement each other: meaning-based search gets you to the right person fast, and the connections around them tell you who else is relevant. Paired with a relationship timeline for each person, you can move from “who knows supply chain?” to the full thread of your history with them in a couple of taps.

Key takeaway: Semantic search finds people and notes by meaning instead of exact words, which matches how you actually recall — but for relationship notes it must search only your own saved notes, cite its source, and stay private on your device.

FAQ

What’s the difference between semantic and keyword search?

Keyword search matches the exact words in your notes, so it misses synonyms and paraphrases. Semantic search matches meaning, so a query like “climate” can surface a note that says “renewables fund” without any shared words.

Does semantic search read the whole internet to answer my question?

It shouldn’t, for relationship notes. The point is to search your own recorded observations. Intriq’s search runs over the notes you saved and cites them, rather than pulling in web or third-party data about a person.

What happens if I ask about something I never wrote down?

A grounded tool tells you it’s not in your notes. Intriq won’t invent a detail to satisfy the query; it answers only from what you actually recorded and is upfront when there’s nothing there.

The payoff

Semantic search turns a pile of hurried notes into something you can actually interrogate by meaning, the way memory is supposed to work. Kept grounded in your own notes and private on your device, it makes recall before any meeting far less dependent on remembering the exact words you once chose. Intriq builds search and recall on that principle. For the bigger picture of how grounded AI should behave, see how AI recall should work and the AI relationship assistant hub.