What's New: AI Task Builder enhancements, expanded participant pools, and more
Welcome to our November 2025 product updates!
This month's releases give you more flexibility in how you design and run AI training and evaluation projects, while expanding access to specialist participants worldwide. Whether you're collecting annotation data at scale or recruiting niche expert groups, these updates help you work faster without compromising on quality.
Duplicate AI Task Builder batches to speed up workflow
You can now run multiple rounds of AI annotation tasks faster than ever. Duplicate batches in AI Task Builder to eliminate repetitive setup work when you need to collect additional data or test new instructions.
- Duplicate everything (dataset, instructions, and task details) to collect a fresh set of responses from new participants on the same data. Perfect when you need more annotations for the same dataset.
- Duplicate settings only (instructions and task details) to apply your carefully crafted instructions and configurations to completely new datasets without rebuilding from scratch. Ideal when you're running the same annotation task across multiple data batches.
This means less time on manual setup and more time analyzing results. Whether you're iterating on evaluation templates or scaling up successful annotation tasks, batch duplication lets you move faster.
AI Task Builder now supports quota and representative samples
You can now use quota and representative samples for AI Task Builder studies, giving you more precise control over participant demographics for your AI training and evaluation tasks.
- Representative samples: Target participants based on UK or US census data, including age and gender distributions, political affiliation, and regional stratification that mirrors population density.
- Quota samples: Set specific participant distributions across demographics. For example, recruit a 50/50 split of male and female participants, or target 30% from Spain and 70% from France.
This helps you ensure that your data reflects the exact demographic mix you need. Learn more: representative samples, quota samples, and AI Task Builder.
New specialist participants now available
Finding specialized participants for research or AI development shouldn’t slow down your projects. This month, we've expanded our participant pool to include hard-to-reach groups across different areas.
- Medical professionals for healthcare AI and research: We've added 200 physicians across multiple specialties, including GPs, psychiatrists, and emergency medicine specialists. Ideal for evaluating healthcare AI outputs or conducting clinical research.
- Audio specialists for voice AI and speech models: Our new pool of audio annotation and generation experts brings hands-on experience with speech and sound data, supporting everything from voice assistant training to audio content evaluation.
- Global coverage for diverse perspectives: Over 1,000 active participants across Asia-Pacific, plus expanded coverage in Middle East, North Africa, and China expat communities. Access a broader range of cultural and linguistic perspectives, essential for globally representative research and AI systems.
These specialists join our existing network of 11,000+ AI-annotation experienced participants and 3,000+ Domain Experts. Read more about participants skilled in AI tasks.
How Prolific delivers trusted data quality: Understanding Protocol
Your research is only as good as your data. That's why we never stop upgrading Protocol: our multi-layered defence system that protects your research from bots, AI-generated responses, and fraudulent participants.
From identity verification to real-time behavioural analysis, Protocol's 5 powerful layers work together to ensure every response is authentic. Watch our 90-second video to see how it works, or read our article to learn more about each layer.
