A connection, not an install
No new software, no new account.
Just connect MCP to the AI you already use.
vling MCP|YouTube data, wired straight into your AI
Looking up channels one by one and copying the numbers across
now takes one message to the AI you already use.
Two minutes to connect. Nothing to install.
vling MCP demo conversation
Find beauty channels over 100K subscribers with a calm, understated tone
Found 12 channels that match. Here are the top 4 by average views.
Channel names masked for the example · every figure comes from vling data
An AI only knows what it can find on the internet. Connect MCP and it can use your own data as well.
Each AI tool knows something different.Every new tool means supplying the material again.
One MCP covers all of them.Whichever AI you use, it sees the same data.
No new software, no new account.
Just connect MCP to the AI you already use.
You don’t have to spell out which data you need.
Say what you want,
and the AI works out exactly what to pull.
It makes no difference which AI you use.
If it supports MCP, it connects the same way,
and the connection stays once it is made.
Connect vling MCP and the AI also sees vling’s own data, the kind search never reaches, so the answers come back better.
Search YouTube and you get subscribers and views.
Build a channel list with AI and you hit the same wall every time.
A few famous channels and years-old numbers. With vling MCP connected, the AI pulls channels that match, with yesterday’s numbers.
“Recommend beauty channels with 300K subscribers”
A few famous channel names. The figures are years old, or simply wrong.
18 beauty channels at 280–320K · by average views
“I cannot verify exact data,” or a plausible guess. Once connected, it shows the actual gender and age split.
“What is this channel’s audience gender split?”
“I cannot verify exact data,” or a plausible-sounding guess.
There is nothing to compare against. Once connected, it lines the channel up against peers of the same size.
“Check whether this channel’s rate is fair”
Generalities. No comparison against channels of the same size.
The problem is not how good the AI is. It is that the data was never connected.
AI answers from knowledge frozen at training time and fills the gaps with something that sounds right. That is why none of it can go into a proposal.
MCP is the standard that links AI to outside data. Like USB-C, where one cable fits everything, you plug it in once and it stays.
Todayfive steps by hand
Connectedtwo steps and done
“Analyze this channel for me”
The copying step disappears. Looking things up and writing them up happen in one conversation.
vling · 12 channels found
241K
186K
Answers come only from the channel and video data
vling collects and cleans,
so they can go straight into a proposal.
Audience
F 20s46%
F 30s28%
M 20s12%
Not mention counts or keyword ranks,
but subscribers, views, audience age and gender,
estimated rates, and sponsorship history.
Proposal draft
Shortlist
Compare metrics
Writing…
Search by criteria to find candidates,
verify with comparison and benchmarks,
then draft the proposal in the same place.
AT A GLANCE
Connect once, keep using itJust a server address and an access key.
Connect in the AI you already use.
Channel search and analysis,
video search, brand data
No signup, no install.
We organize the data collected every day by the criteria you want, and deliver only the results you need through the API.
Address and key, once. After that, you just ask.
Open Connectors in Settings
and add a custom connector.
Enter the name and URL,
then put the key in the request header.
One line in the terminal.
claude mcp add --transport http vling \ https://mcp.vling.net/mcp
Open Connectors in Settings, add a custom MCP,
set the auth type to Bearer token and enter the key.
vling MCP pulled the data and the AI organized it on the spot. Nobody copied a figure across or drew a chart by hand.
01Four beauty brands compared
“Compare four beauty brands on YouTube mentions over the last 30 days. Lay out total views, video count, and average views per video, then split the views into positive, neutral, and negative as a scorecard.”
Total views, video count, and views per video side by side,
with the same views split again by sentiment.
02Sponsorship vs. negative sentiment
“Compare these four brands on sponsored share and negative share. Put sponsored share of mention videos on x, negative share of views on y, build a quadrant chart, and read out how the strategies differ.”
Crossing the two separates brands buying their way past negativity
from brands spending little and still well liked.
03Rival clubs, mentions over time
“Compare two clubs on recent YouTube mentions. Stack daily mention views by sentiment to show the trend, then explain the spikes down to the spoken line and its timecode.”
It catches a single day outscoring the rival’s entire 102-day window,
and traces the cause back to what was said on camera.
04Inside the mention videos
“Analyze the videos that mention us, ranked by views. Lay out the channel type and performance of the top videos, then compare against our own best video to show the reach outside channels added.”
The top slots went to late-night TV, finance, and gaming, not sports,
reaching 3.8× the club’s own best upload.
Describe the criteria and candidates come back. Ask for evidence and a comparison comes back. The proposal draft happens in that same conversation.
Describe the brand tone in plain words. Once the list is back, narrow it on the spot.
“Find beauty channels over 100K subscribers with a calm tone”
12 channels match the criteria
See where a channel sits against its peers. Rates are estimates, and the response says so.
“Compare these two channels on subscribers, rate, and gender”
Check which channels a rival worked with, how, and how it landed.
“Show me the best-performing sponsored videos recently”
Connect Notion too and the results land straight on a page.
“Build a casting proposal table from those five channels”
The output holds its shape even when the owner changes.
“Same format as last month. Write up this month’s status.”
Questions you can paste in right now
You never pick a tool. Ask the way you normally would, and the AI decides on its own whether it needs vling data.
Find this channel for me (name, @handle, or YouTube URL all work)
Find camping and outdoor channels, 50K to 200K subscribers, skewing male
Write up this channel’s recent growth and core audience
Build a table comparing these three on subscribers, average views, and estimated rate
How does this channel perform against others its size?
Show the ten most-viewed videos mentioning our brand in the last three months
Check my usage (today’s usage, remaining calls, reset time)
Channels known by a nickname may not come up. Add the YouTube URL or @handle and it will.
We are running a free trial for early customers. No payment. Get an access key and the whole team can start.
Trial terms at a glance
For early-access customersThe trial costs nothing.
Extended on request.
Raised right away if you need more.
“Check my usage”
Check what is left in the chat.
Each channel you look up or compare counts as one call.
Support actions like checking usage do not count.
Not per seat. The team shares one allowance,
so adding people does not add cost on its own.
Tiers follow how many channels and calls you use.
Light use stays light.
Bearer, stray spaces or line breaks around the key, and the vmk_ prefix.
Saving the settings again and starting a fresh conversation often helps too.No create, update, or delete tools are offered at all.
There is no path for the AI to change anything.
The server stores an encrypted value, never the key itself.
Usage is recorded per key and per team,
so anything unusual can be acted on immediately.
Thirty minutes is enough.
We will show you which of your current tasks turn into a single question.