Search inside your videos and ask your archive

echoflow keeps your videos queryable: with AI Chat you ask a plain-language question and get an answer with its sources: the video and the minute it comes from — even without remembering the exact words. It's the part of the product that gains value over time instead of wearing out.

One question, an answer with its sources

The written answer comes from your content: "what's the conclusion?", "what did I say about pricing?", "summarise the middle section". Underneath are its sources: the videos it comes from and the minute, which open the video at the right point. You can leave it free across the whole archive or narrow it to the videos you pick. It answers from your videos, not from the internet.

Passages by meaning — the exact moments across all your videos where you talked about it, with minute and link, no rewritten sentences — are searched from the MCP connector or the API: they're for when you don't know where to look. They search by meaning, not by word — the difference between finding and not finding something you said two years ago in words you no longer remember.

Inside a single video's row on the dashboard there are still two separate commands: "Search inside" finds a word in that video, "Ask AI" asks questions about that video only.

Why searching by meaning matters

A text search only finds what you wrote identically. Search for "how to handle difficult clients" and a word-based search won't find the passage where you said "when someone complains, the first thing to do is listen" — which is exactly the piece you needed.

Across ten videos the difference is negligible. Across two hundred, it's the difference between having an archive and having a folder of files.

Isolation: not a screen, the database

In an agency with several clients the right question is whether one client's content can show up in another's search. It can't. The index is isolated per workspace at database level (row-level security): a search can't cross two workspaces even through a programming mistake. The same holds for search, chat and Brand Voice examples.

What we don't do

• We don't guarantee exact answers. They're grounded in your content — far better than a generic

model — but an ambiguous passage or an imprecise transcript produces an imprecise answer. Verify before quoting them to a client.

• Chat doesn't browse the web and knows nothing beyond your content: if the answer isn't in your

videos, it won't invent one from elsewhere.

• We don't search inside anyone else's videos: only yours, only in your workspace.
• Indexed content is text, so it doesn't expire: the archive stays queryable even after the

seven-day media files are gone.

Which plan

AI Chat is included in the Pro plan (from €19/month). Answers draw on the monthly AI tools allowance included in your plan — a single, shared one: roughly, it's worth around 600 AI Chat messages a month, in both sizes. It resets on the first of the month and doesn't consume minutes. Passage search by meaning, with minute and link, is used from the MCP connector or the API, also on Pro, and doesn't use the allowance.

The archive is queryable from outside too: with the MCP connector you can search your videos directly from Claude or Codex, with the same thresholds as the web app; with an API key, also on the Pro plan.