|
05-Oct-2026 |
Hindsight Structured Advice Distillation: Teaching a Small Model to Guide a Research Agent. This research fine-tunes a small advisor on an agent's successes and failures to write question-specific advice. This approach boosts a research agent's performance on the FrontierFinance leaderboard without retraining. |
Research / Article |
|
28-Sep-2026 |
FrontierFinance benchmark: Financial research measured against expert judgment. This page details the FrontierFinance benchmark, an open benchmark for frontier financial intelligence that measures how well AI systems support the full investment workflow. It includes system quality vs. cost comparison and performance breakdown. |
New Product |
|
11-Sep-2026 |
Samaya's fifty workflows are designed for financial analysts, covering recurring and unexpected tasks. They handle data gathering, mapping, and analysis, allowing analysts to focus on judgment, clients, and primary research. |
Blog |
|
09-Jul-2026 |
Introducing FrontierFinance: A Challenging Benchmark for Measuring Frontier Intelligence of Finance Agents. We release FrontierFinance, an open benchmark for measuring how well finance AI agents perform across the full investor workflow. It contains 220 expert-crafted queries spanning six finance use cases, making it the largest open benchmark of its kind. |
Research / Article |
|
23-Jun-2026 |
Before the Search: Why Query Understanding Matters for Financial Agents. This article explains why query understanding is a critical challenge for AI agents in finance. It details how Samaya trains models for this purpose and the resulting improvements in performance. |
Research / Article |
|
27-May-2026 |
Samaya AI's ToolObserver lets agents learn tool behavior from actual task trajectories instead of static docs. On the new OpaqueToolsBench, it beats state-of-the-art methods by 18.6% while using 3.5–7.5× fewer tokens. |
Research / Article |
|
27-May-2026 |
When Tool Documentation Falls Short: Teaching Agents to Learn From Interaction. Samaya AI's ToolObserver allows agents to learn tool behavior from actual task trajectories instead of static documentation. It reportedly beats state-of-the-art methods on the OpaqueToolsBench. |
New Product |
|
16-Feb-2026 |
Samaya AI researchers presented 'OpaqueToolsBench: Learning Nuances of Tool Behavior Through Interaction' at ACM CAIS 2026. This research focuses on understanding tool behavior through interaction. |
Research / Article |
|
10-Feb-2026 |
Samaya AI launches Agent Control Plane (ACP) for personalized finance agents and secures new investment from NVentures and Databricks Ventures. |
Funding,New Product |
|
08-Jan-2026 |
Samaya partners with Third Bridge to bring expert insights directly into AI-powered investment research. By integrating Third Bridge’s premium expert interviews directly into Samaya’s Expert Intelligence Platform, investment professionals can seamlessly access trusted, qualitative insights. |
Partnerships,Integrations |
