Know every respondent is a real, continuous human. Before they enter your panel.
Bots, survey farms, and duplicate accounts sit inside online research recruitment today. Pupul gives each participant a portable, consented record of real human activity, and answers one question about them at intake: is a continuous human standing behind this response.
15 to 45%
of respondents in some online studies are estimated to be fraudulent, duplicate, or bot participants, the range researchers have reported as data-quality threats in online recruitment. Griffin et al. and others, online-panel fraud literature.
How it works
You do not collect anything the participant wrote, and you do not run a test at the door that a better model will eventually pass. You read a record the person has been building, with their consent, and revoke-able by them in two clicks.
01 · The person
A participant carries a Pupul record, a portable and consented history of real human activity that they own, not you.
02 · The check
At intake, one call over REST or MCP returns four claims: is human, is continuous, has deviated, stands behind. A claim we cannot evidence returns null, never false.
03 · The record
The proof is anchored on a public, append-only transparency log. Elapsed, consented time cannot be backfilled, which is the one thing a model cannot fake.
Plans
Every plan includes the four claims over REST and MCP, participant-owned consent and revocation, and access to the public transparency log. Monthly, cancel any time.
Solo
$99/mo
An independent researcher or a single study verifying participants.
Prices are in US dollars and billed monthly. The four claims are the same on every plan; the plans differ in volume and support. Need a custom volume or a signed agreement first? Start with a pilot below.
Run a pilot first
If you would rather prove it on one study before you subscribe, we run short paid pilots with research teams: your intake, your panel, our check, measured against your own duplicate and fraud rate. You keep the data quality lift whether or not you continue.
No. Pupul answers claims about whether a continuous human stands behind a response. It is not handed your survey content, and the participant, not you, owns and can revoke the record.
What if a participant has no record yet?
Then the claim returns null, never false. An absent record is not an accusation, and you decide how to treat a null at intake.
Why is elapsed time the point?
Every test given at the door, a puzzle, a face, a gesture, is a race against the model that will eventually pass it. A consented record that has been accruing on an append-only ledger the whole time cannot be minted after the fact. See proof of human.