Peptide Signal
We are building an API and data product for research teams and industry: structured early-adopter demand and outcome signal for peptide therapeutics, drawn from what people actually post while using them — months before the same compounds reach a trial readout or a label.
What exists today
These are counts from our production dataset, not projections. The corpus has been ingested and classified continuously since 7 June 2026 and grows daily.
- 5,399
- classified posts2,700 Reddit · 2,699 X
- 33
- compounds trackedwith alias matching
- 19
- source communitiesplus continuous X ingestion
- 7
- classification clustersexperience, dosing, side-effect…
What you would actually get
Post volume is not the product — anyone can count mentions. The value is in a domain-literate classification layer on top, which turns unstructured discussion into figures you can compare across compounds and over time.
Outcome verdicts, with denominators
What share of people reaching a conclusion said a compound worked, split by what they were trying to achieve — and how many that share is out of. A 100% success rate from five people is reported as exactly that.
Discontinuation rates and stated reasons
What proportion stopped, and whether they said it was side effects, cost, reaching their goal, or no effect. This is the figure almost never quoted alongside a success rate, and it frequently points the other way.
Self-reported outcome curves
Median reported change by week-bucket, with sample size and spread at each point. Enough structure to see where a reported trajectory flattens.
Time-to-verdict
How long people actually run something before concluding anything. It ranges from under a week to ten weeks across the set, and it correlates with how little evidence a compound has.
Side-effect frequency tables
Symptom, count, and share of posts, per compound. Self-reported and self-selected — stated as such, never dressed up as incidence.
Supply-channel split
Pharmacy, compounded, research vendor, or unstated — per compound. The split tracks approval status almost perfectly, and it is the figure that tells you whether anything else in a dataset can be trusted.
Open datasets, free today
Aggregate cuts of the corpus, published openly under CC BY 4.0 with a CSV download. These are the shape of the thing — the commercial product goes deeper, but these are real, current, and yours to use.
Peptide discussion volume and sentiment by compound
How much each compound is discussed, and how positively.
View + CSV →What people discuss, by compound and topic
The share of discussion that is dosing, side effects, sourcing and so on.
View + CSV →Side-effect discussion intensity by compound
Which compounds generate the most tolerability discussion, and how negatively.
View + CSV →Also free: the development pipeline
Every peptide and incretin trial we track — sponsor, phase, status and expected readout, from public registry records. Open the pipeline tracker.
Who it is for
- Competitive intelligence and market insight teams tracking how pipeline compounds in obesity, metabolic and recovery indications are discussed before launch — including the grey-market demand for a molecule that precedes its approval.
- Medical affairs and patient-insight teams who need to know how real users describe tolerability, adherence and the reasons they stop, in language trial populations do not use.
- Equity research and investors looking for demand signal ahead of readouts and launches.
- Academic researchers studying self-medication, health misinformation, or unregulated supply. Terms for non-commercial academic use will be separate.
What this product does not do
We do not sell post content. The product is derived statistics — counts, rates, distributions and curves. We do not redistribute post bodies or author identities, and aggregates below a minimum sample size are suppressed rather than reported, because a figure drawn from four people is not a figure.
We do not perform adverse-event case identification. Side-effect data is supplied as aggregate frequencies for market and tolerability analysis. It is not pharmacovigilance intake, it is not structured for individual case safety reports, and it is explicitly out of scope — a choice we have made deliberately rather than one we are hedging.
Self-reported data is not incidence data. Everyone in this corpus chose to post. People who experience nothing rarely write about it. Every figure we ship carries its denominator and its sampling caveat, because a rate without them invites exactly the wrong conclusion.
The editorial firewall
Our published research is why this data is worth anything, so it is not for sale. Data customers get no input into editorial coverage, no advance sight of articles, and no influence over how any compound is graded in our evidence-tier framework. If a customer’s compound sits at tier 5 because it has no human data, it stays at tier 5.
We also do not sell to vendors we cover or grade. Our editorial policy sets out the rest.
Join the waitlist
Tell us where to reach you and we will be in touch as access opens. We would rather talk to a small number of teams about what they actually need than ship a generic feed — so expect a conversation, not a drip sequence.
Pricing is not published while the product is in development. For partnership, licensing or academic enquiries, say so when you sign up and we will route it accordingly.
Counts on this page were taken from production on 19 October 2026 and describe the corpus at that date. Nothing here is medical advice, and the dataset describes what people posted — which is not the same as what is true. That distinction is the whole point of the product.