Deel is adding 'Pangaea' to 'Deel AI,' using two specialist language models to answer country-specific leave, employment, and pay questions for HR teams and employers across jurisdictions. 'K1' covers work rules, 'P1' handles payroll, and 'Emet' provides the expert training material and evaluations behind them.
K1 began as a 9-billion-parameter model and was retrained on Deel's reference library, cutting how often the assistant has to search documents before replying. In Deel's benchmark, it posted a 7.7-second median response time, giving employers faster guidance before a payroll error reaches a paycheck.
A 7.7-second median response time makes specialist models easier to justify in subscription HR software, where external AI calls add cost. Emet turns the same payroll and compliance know-how into a B2B product for AI developers working in regulated categories.
Image Credit: Deel
What's Driving This Trend
- Specialized Compliance Models
- Domain-trained AI assistants create space for faster, jurisdiction-aware HR guidance that reduces dependence on broad general-purpose models.
- Real-time Payroll Intelligence
- Low-latency answers about pay rules and employment requirements support new software experiences that flag compliance issues before payroll is finalized.
- Regulatory Knowledge Products
- Expert-validated training libraries and evaluations are emerging as commercial infrastructure for AI builders serving highly regulated business functions.
Who This Affects Most
- Human Resources Software
- HR platforms can differentiate through embedded compliance intelligence that helps employers manage leave, work rules, and payroll across countries.
- Payroll Services
- Payroll providers face new opportunities around automated rule interpretation that lowers error risk in multi-jurisdiction workforce operations.
- Regtech
- Regulatory technology vendors can expand beyond monitoring into model training, benchmarking, and expert data products for enterprise AI systems.
