A platform designed to elevate work

We view work much like a pyramid. The highest leverage work is at the top of the pyramid. The most time consuming work is at the bottom of the pyramid. Many see agents as a threat to human work. We see human knowledge work as dominated by repetitive, mundane tasks. Our ambition is to elevate work by leveraging agents at the bottom of the work pyramid.

Oversee

As agents and AI systems take over human work, it is critical that human experts act as a final arbiter of policies. We are committed to full transparency. Ironically AI work brings far better metrics than human work.

Human Oversight

Humans ultimately oversee all work performed by agents and algorithms.

Agent Metrics

All agent work is measured on a pass/fail basis for criteria such as relevance, safety, accuracy, financial clarity, tool reliance, etc.

Performance Dashboards

Dashboards display performance of agents and algorithms.

Match and Resolve

We have developed foundational agent graphs designed and measured to perform deep research on financial transactions. These foundational features are based on a matching algorithm that enables agents to read and match the root cause explanations of past resolved issues.

Agent Workflows

OnCorps AI has developed foundational agent graphs designed to perform detailed research work.

Matching Root Causes

Agents call on matching cases of past resolved exceptions to improve confidence in root causes.

Position-Level AI

Algorithms track individual positions across their full lifecycle. Day-over-day changes are measured for anomolous patterns.

Detect

We have dedicated significant time and investment in training algorithms on financial positions. Our work with the leading asset managers has resulted in algorithms that are fine-tuned with expert feedback from our leading customers. These algorithms can be shared through a sophisticated weights-only binding method.

Fine-Tuned Algorithms

Bottoms-up position level tracking helps detect anomalies through rigorous isolation and sampling.

Focused Checks

Incident detection and other functional checks are maintained and deployed to identify patterns that contributed to past errors.

Shared Algorithms

We can share pre-trained algorithms securely. Only the weights are shared with no proprietary data mixed.

Read

We have developed features to ingest data from many sources, cleanse it, and curate it to feed our algorithms. We also have developed a partnership with Xceptor to speed data management features.

Data Quality

Rigorous adherence to a staged medallion architecture. Bronze layer is initial staging of all IBOR, ABOR, and third party data. Silver layer is cleansed and confirmed data, and gold layer is curated tables ready for algorithm use.

Secure Platform