Demo — Conversational AI and agents
Ask the site.
Questions about Friendify, answered with what this site actually says, quoted and linked. The search runs in this page: your questions aren’t sent anywhere.
Checking what this device can run…
How it works
Search by meaning, answer with the site’s own words.
A question goes in; the passages that answer it come out, quoted and linked. Nothing is written by a model, so nothing is made up.
You ask
Where is the company based?
Step 1
Read the site
When the demo loads, the site’s pages are split into about 90 short passages.
90 passages, from capabilities to the privacy policy
Once, on loading
Step 2
Turn meaning into numbers
A small open model turns each passage, and then your question, into 384 numbers.
Your question, as a point in meaning
bge-small · 35 MB
Step 3
Find the nearest
Passages that mean something close to the question come out close.
Company information · match 0.62
About 0.1 s
Step 4
Quote, with the source
The closest sentence is marked, and links to the page it came from.
“Registered office: …, London”
Nothing generated
The answer
The passage itself, its page, and the sentence that answers.
- downloaded once with the engine that runs it, then kept by your browser
- 62 MBdownloaded once with the engine that runs it, then kept by your browser
- to answer a question
- 0.1 sto answer a question
- words generated: only quotes
- 0words generated: only quotes
For clients
Your documents, the same way.
This is the search behind the support and sales assistants we build: answers grounded in your own documents, with sources a person can check. For a client we index their help centre, policies or product data, keep it in step as it changes, and measure how often the right passage comes first.
Where an assistant should answer in its own words, a language model writes the answer from these passages and shows them underneath, so every claim can be traced.
About this demo
Small on purpose.
The demos here download 50 MB to 1.0 GB: small enough to fit into a web page and run on your own device, with nothing sent to us or to an AI company. Your browser keeps the files, so a second visit starts from that copy.
They do a great deal for their size, and they make mistakes a larger model wouldn’t. What we build for clients uses larger models where they’re needed, the client’s own data and words, testing before launch, and a person to hand over to.
The files stay in your browser until you remove them, or the browser needs the space. Removing them here clears every demo’s models from this device; a demo downloads its own again if you load it.