prompt & get paidOpen workspace ↗AI needs more than compute.
TECHCRUNCH · SEPTEMBER 2026AfterQuery reportedly becomes YC’s fastest-ever unicorn.Reported valuation: $3.2 billion.Read the reporting ↗Reporting about AfterQuery, an independent company.
It needs people who can spot a wrong answer, explain a better one, and bring knowledge a model is missing.
AfterQuery, Scale, and Kled have built businesses around data and human judgment. That is the opportunity we are working toward.
People also pay for access. Claude Max 20x is $200 a month on the web. Cursor Pro+ is $60. Some consumer services use conversations for training when the applicable settings and consent allow it.
Claude plansCursor plansOpenAI data choicesAnthropic data choices
We think useful contributions should give something back to the person making them.
Start with a good workspace.
Three included text prompts let you try it. Sign in with email to save conversations and use the catalog with your available credit.
Chat models have clear names and use cases. Image creation has a separate model picker, sensible default sizing, saved results, and full size downloads. Ask for a portrait or another shape in the prompt when you need one.
Prompt polish rewrites a draft for the model you chose. You review the change before sending. You can always keep the original.
One small task. A useful record.
Earn is built for a choice and a short reason. The record behind that choice keeps the task, its source, the scoring rules, and your permission together.
Quality checks and review decide which work is accepted. Repeated clicks are not the goal. A specific observation is.
Our aim is to collect useful judgments across languages and disciplines, then build datasets with a clear purpose and evidence of quality. A task count alone is not a finished dataset.
Your chat. Your choice.
We never sell or license your private chats. Only evaluations you choose to submit can enter licensed datasets. Saved conversations stay in your history. Every submission shows what you are sharing and asks for permission.
You retain ownership. The displayed terms allow accepted contributions to be used and sublicensed for AI evaluation, training, and related datasets. Only submit work you have the right to share.
Tasks keep source and consent records. Public references do not imply a company commissioned the work. Any dataset offered to buyers needs its own quality and rights review.
Credits pay for usage.
Models cost different amounts. The app reserves credit before a request, charges for usage afterward, and returns any unused reserve.
Sponsored usage and Earn rewards have funded limits. Credits cannot be spent twice, and a failed request releases its reservation.
Current rewards are product credit. Accepted work adds credit for AI inside the app. It cannot be withdrawn as cash or tokens. Each task shows its possible reward before you submit.
Usage revenue, sponsorship, and future dataset sales can help fund the experience. Dataset sales are a goal, not a promised source of income.
Token utility, when Phase 2 opens.
The token is planned as an access and participation layer around the product. It is not a promise of profit, a fixed credit grant, or a claim on company revenue.
Hold. Verify. Unlock.
When the holder program is activated, a wallet holding the published threshold of the production token can sign a verification message and unlock selected premium models and Prompt polish. The draft launch policy uses about $50 of token value for entry and about $40 for continued access. The final threshold can change before activation.
Collected fees fund the loop.
The planning model first reserves 10% for collection, execution, and uncertainty. The remaining usable fees are modeled as 40% AI provider funding, 20% future Earn or payout funding, 30% operations, and 10% reserve. These are planning allocations, not live automated transfers.
Credits follow real economics.
If the program is activated, new retail credits are modeled as 50% for active eligible holders, 30% for accepted Earn work, and 20% for acquisition. Any holder allocation depends on collected fees, active participants, provider costs, and treasury limits. It is capped and never represented as cash or USDC.
Not live yet. The holder gate, token address, creator-fee collector, buyback or burn, and USDC payouts are disabled. Product credits remain separate from any future payout program. No fixed return is promised.
Start with AI. Build the value loop.
One workspace, growing in three phases.
Prompt. Create. Earn AI.
Leading chat models, image creation, and optional evaluations. Accepted work earns product credit you can use in the workspace.
Turn useful work into payouts.
Wallet-linked access and USDC payouts for eligible accepted contributions. Rewards will come from a funded pool, with clear review and payout rules.
Product credits remain separate from withdrawable rewards. Existing credits will not automatically convert to USDC.
More work. More ways to earn.
Specialist evaluations, computer use tasks, video creation, and datasets built around buyer demand.
Also planned: a private mode that routes supported models only to inference endpoints with a verified zero data retention policy.
Future features are planned, with no fixed release date. USDC payouts depend on funding, eligibility, security, and applicable requirements. Private mode will explain saved history and tool policies separately. Token ownership will not make AI usage unlimited.