firestore Vector Save of, "what the best bet...", matches retrieval for, "foo"

Either I've set this up wrong or I'm misunderstanding the vector database.

I have the phrase, "What's the best bet on the table", I'm embedding it to get the vector, then savings to firestore as a vector. My intention is to match sentences like, "What's the BEST bet on the table" with a phrase like, "What's the TOP bet on the table".

Here is my save code:

require('dotenv').config({ path: __dirname + '/.env' })

const { initializeApp, cert } = require('firebase-admin/app');
const { getFirestore, FieldValue } = require('firebase-admin/firestore');
const serviceAccount = require('./secrets/crapsai-72b89-firebase-adminsdk-pnqva-e321c8eecb.json');

const { gemini15Flash, googleAI } = require('@genkit-ai/googleai');
const { textEmbeddingGecko001 } = require('@genkit-ai/googleai');

const { genkit } = require('genkit');

const ai = genkit({
    plugins: [
        googleAI({
            apiKey: process.env.GOOGLE_GENAI_API_KEY
        })
    ],
    model: gemini15Flash
});

const app = initializeApp({
    credential: cert(serviceAccount)
});

const firestore = getFirestore(app);

var question = "What's the best bet on the table";

(async () => {
    try {

        const embedding = await ai.embed({
            embedder: textEmbeddingGecko001,
            content: question
        });

        console.log("Got the embeddings");
        console.log(embedding);

        await firestore.collection("questions").add({
            "question": question,
            "embedding": FieldValue.vector(embedding),
            "confirmed": true
        });

        console.log("DONE");
        process.exit(0);
    } catch (e) {
        console.log("Error", e);
        process.exit(1);
    };
})();

Here is what that looks like in firestore

Here is the retrieval code:

require('dotenv').config({ path: __dirname + '/.env' })

const { initializeApp, cert } = require('firebase-admin/app');
const { getFirestore } = require('firebase-admin/firestore');
const serviceAccount = require('./secrets/crapsai-72b89-firebase-adminsdk-pnqva-e321c8eecb.json');

const { gemini15Flash, googleAI, textEmbeddingGecko001 } = require('@genkit-ai/googleai');

const { defineFirestoreRetriever } = require('@genkit-ai/firebase');

const { genkit } = require('genkit');

const ai = genkit({
    plugins: [
        googleAI({
            apiKey: process.env.GOOGLE_GENAI_API_KEY
        })
    ],
    model: gemini15Flash
});

const app = initializeApp({
    credential: cert(serviceAccount)
});

const firestore = getFirestore(app);

const retriever = defineFirestoreRetriever(ai, {
    name: 'questionRetriever',
    firestore,
    collection: 'questions',
    contentField: 'question',
    vectorField: 'embedding',
    embedder: textEmbeddingGecko001,
    distanceMeasure: 'COSINE'
});

var question = "foo";

(async () => {
    try {
        const docs = await ai.retrieve({
            retriever,
            query: question,
            options: {
                limit: 5,
                where: { confirmed: true }
            },
        });

        docs.forEach(doc => {
            console.log(doc);
        })

        console.log("DONE");
        process.exit(0);
    } catch (e) {
        console.log("Error", e);
        process.exit(1);
    };
})();

Then when I run the code for retrieval based on the question, 'foo' expecting to NOT get a match. I get a match.

SO either I've set this up wrong or I'm misunderstanding the vector db.

Here is my index in firestore

SOLVED: So it's ranking them, I had so little data it was ranking the unliked highest. When I added more data I could see that it's working, it just returns something no matter what.

The code above in the post worked for me.