live now · shipping weekly

free invisible bot protection.

Panel is a Turnstile-like verification layer that stays invisible for trusted humans and escalates only when risk is unclear. Every verification produces compact abuse signals that train the edge model, so checks get quieter as the network learns.

L1 live · running now
the ladder

invisible by default, useful when it escalates.

Most visitors pass through on behavioral and environment signals alone. When risk is unclear, Panel can step up to human-safe judgment tasks. The optional paid loop routes expert/agent-output reviews to trusted raters.

L1 · invisible check
replaces recaptcha & turnstile.

Panel starts with passive behavioral and environment checks. Clean traffic gets a token without a puzzle. Ambiguous traffic escalates to short, grandma-safe tasks that also create model-training labels.

L2 · agent-output rating
b2b feedback loop for agent stacks.

operators emit agent outputs, skill diffs, process outputs. trusted raters 'judge' them — pairwise, rubric, free-form. preference rows flow back to the operator, signed, deduped, scored. the loop closes without a labelling vendor in the middle.

L3 · expert review
domain specialists for high-stakes signal.

regulated pros, medical, legal, trades. credentialed reviewers gated behind verification. paid per unit, audit trail per row. for outputs where a wrong rating costs more than the rating.

the wedge

turnstile gives you a token. panel gives you a learning edge.

The free product is bot protection, not a data-labelling ask. The flywheel is simple: verification traffic hardens the edge model; paid operators can opt into panel-data and rater workflows later.

recaptcha · turnstile

visitors label google's self-driving data for free.

operator gets a yes/no token. the dataset goes to mountain view.

rater compensation: zero.

panel

visitors judge the operator's own agent output.

operator gets a yes/no token and the labelled row.

rater compensation: balance accrues. paid-train coming.

emitter pipeline

three lines. no hmac dance.

the sdk handles signing, batching, and the ingest contract. operators emit. adapters fan out. raters judge. signal returns.

panel-sdk · typescript
import { createClient } from 'panel-sdk';

const panel = createClient({ siteKey: process.env.PANEL_KEY!, secret: process.env.PANEL_SECRET! });
await panel.emitProcessOutput({ kind: 'reply', content: agentReply, context: prompt });
shipping
modal dev-gen

serverless agent runners emit process outputs straight from the function. hmac-signed in the worker, no proxy.

incoming · v6b
comfyui · replicate

image/video generation pipelines. webhook adapter writes media_origin + media_quality units on completion.

incoming · v6c
elevenlabs

tts/voice clone outputs into dub-sync and naturalness rating units. same emit shape, different rubric.

paid-train loop

rater balance → tokens → finetune → deploy.

raters earn by judging. balances convert to inference credits and finetune runs against the dataset they helped label. operators close the loop on the same surface that produced it — judge, train, deploy, judge again. no data-broker, no labelling vendor, no separate gpu account.

where it is today

shipped now

L1 live · /embedpersistent sqlitetrust-pool routingbehavioral signalshoneypot unitsoperator-key authpanel-sdk · tsmodal dev-gen adapterno paid-train yetno SOC 2
try it

live surfaces