Most investors never monitor what happens on Instagram.
That is a mistake.
For anything that reaches consumers before it reaches a P&L, Instagram is the best alternative data source available and almost nobody on the buy side is touching it.
Consumers talk in real time their experiences, sentiment towards a product, and more.
I have kept that in my back pocket for a couple of months, waiting for a use case worth spending real time on.
My brother handed me one. He started buying peptides from gray-market dealers online, and he was getting results.
For background .... These are research compounds sold with no approval and no pharmacopeia, and the claims attached to them are enormous: heal a tendon that has been bad for years, sleep like you did at twenty, sharper focus, more muscle, a tan without the sun.
So my question(s) as an investor .... What's the full menu of peptides out there? Which peptides are legit? What are the reported side effects? What are the downstream impacts on consumer products once these are approved by the FDA?
We already know how this movie goes.
GLP-1s started as a diabetes drug, went mainstream, and then repriced whole consumer categories that had nothing to do with diabetes, from packaged food and alcohol to apparel, aesthetics and sleep apnea devices.
Some influencers like Derek from More Plates More Dates have been talking about the science of GLP1's for years prior given they were our human guinea pigs.
That playbook is the reason to care about a bunch of unapproved compounds being sold on Instagram.
If even one or two of them clear approval, the same cascade runs again.
What we can do with Instagram (combined with other consumer MCP connectors and scrapers like for Tiktok and Reddit) is understand the full menu of popular peptides that are being sold, what people are willing to pay
Scraping sellers tell you what is selling and what people will pay for. Users tell you, in public and in detail, what it did to them.
Here’s the plan:
Connect the MCP social data layer to your agent
Let it build the census of what sells
Read the comments for effects and side effects
Count mentions over time to find acceleration
How I did this
I use xpoz as the MCP (btw, I’m not paid by them and there are many other Instagram MCPs). It sits between your agent and the social platforms, so you never need a Twitter API key or an Instagram scraper of your own.
I set this up in the Claude desktop app. Here's how to connect in about two minutes.
Step 2. Open the Claude desktop app, go to Settings, then Connectors
Step 3. Click Add custom connector and paste in the xpoz server address: https://mcp.xpoz.ai/mcp
Step 4. Click Connect. Claude walks you through the Google sign-in, and that is what issues your key behind the scenes.
That is the whole setup. xpoz now shows up in your connector list, and Claude can query Instagram, Reddit, TikTok and X directly in any chat.
If you use Claude on the web instead of the desktop app, it is the identical flow. Settings, then Connectors, then the same address.
Once connected, I asked it a prompt that included three things to cover what we're looking for in this peptides research piece.....
The Prompt (steelman)
Full prompt is too long to share here but in essence you can get the same results if you make sure you cover these 3 pillars.
The key to great Claude Code or Codex results is the right "prompt pillars" to steelman your logic:
Prompt Pillar 1, the census. Search Instagram for clinics, med spas and sellers marketing peptides. Build a list of every distinct compound they sell, including ones I have not named. For each, capture the seller type, the claimed benefit, and any price shown.
Prompt Pillar 2, the reviews. For the highest-engagement posts on those compounds, pull the comment threads. Classify each comment as a reported effect, a reported side effect, or a complaint that the product did nothing. Keep the verbatim line and the link.
Prompt Pillar 3, the momentum. For each compound, count mentions across the last ninety days and the ninety days before that, then rank by the ratio. Flag any search term that collides with an unrelated word.
The seller storefront scraping came back with 12 top peptides that grey-zone vendors are pushing:
Of the 42 compounds identified, 28 are running more than 2x prior quarter mention volume on socials.
3 interesting things that things came out of the socials scrape that changed how I look at the theme:
Most of the popular ones have no owner. The compounds with the loudest demand are largely unpatentable and off-patent, so no company has a reason to fund a trial. That does not mean nobody gets paid. It means the money moves to whoever owns the channel instead of the molecule: the compounding pharmacies, the telehealth platforms, and the vendors themselves.
The first-order winners are the molecule owners. Where a real company does own a compound the crowd is already buying, you have gray-market demand acting as a leading indicator for a drug still in trials.
The reviews land years before the label. Efficacy, side effects, dose-response, and even label-expansion hints are all being written in public, for free, and early. For example: A seller account posted PT-141 as "The Desire Peptide" .... Directly underneath, customers: "I took it one time won't touch it again... made me feel like shit.""Horrible peptide I agree. Lasts for about 2 hours and you feel shit."