- Workflows: Autonomously gather, move, and send your data wherever it needs to go. This is the Doer in your toolset, pulling data, performing chains of discrete actions with rich functionality from data operations to integrations and conditional branching.
- Views: Instantly transform, blend, and analyze complex datasets without requiring heavy database engineering. These are the foundations for reports, data transformations, etc. and do the heavy lifting of Data Analysis, Cleanliness, and Formatting.
- Alerts: Provide real-time updates the second something unexpected or out-of-bounds occurs. Keep up to date and get the information you need in real time.
While each of these pillars is powerful on its own, their true magic is unlocked by AI. We aren't just adding AI checkboxes; we are introducing two specialized AI agents designed to handle the heavy lifting of financial data management: the Coordinator Agent and the Error Resolution Worker.
Managing financial data today usually means wrestling with an endless maze of disconnected systems, rigid APIs, and manual spreadsheets. At CoSet, we took a step back and rebuilt this infrastructure from the ground up around three core elements.
From Intent to Execution: A Live Reconciliation Walkthrough
To see how these pillars and agents interact, let’s look at a classic operational challenge: reconciling daily positions between a Prime Broker (PB) and a Fund Administrator (FA).
Instead of writing custom code or building fragile mapping templates, you simply tell the Coordinator Agent what you want to achieve:
"Reconcile yesterday's position data between Apex Prime and Summit Fund Services."
1. Workflows Gather the Raw Data
The Coordinator Agent automatically spins up two Workflows. The first connects to an SFTP server to pull the daily position file, while the second workflow would send a filtered list of breaks to users as errors occur. These workflows ingest the messy, raw formats and stream them directly into clean, structured database tables, with manual sign-offs as desired to validate.
2. Views Transform and Identify Breaks
Once the data lands, the Coordinator Agent stitches the tables together using a View. The View normalizes the disparate data formats, aligns the assets by their identifiers via a predictable SQL Join, and automatically generates a calculated Breaks column. This column instantly flags any discrepancies in quantity, market value, or cash balances between the two sources. These Views do not need specific refreshes or updates, and can be accessed either via the CoSet product or downloaded to csv with filters set in a Workflow.
3. Flag the Exceptions
Finally, a downstream Workflow triggers an email the moment a break is detected. Instead of emailing a massive, noisy spreadsheet to the entire team, the system isolates the specific filtered list of discrepancies and routes it directly to the designated reconciliation owner for immediate review.
Intelligent Resilience: The Error Resolution Worker
In the real world, data pipelines break—often because a human or an external system changes a minor detail. That is where our second agent, the Error Resolution Worker, comes in.
Imagine Summit Fund Services suddenly changes their daily file naming convention from position_report_20260707.csv to Summit_Positions_07_07_2026.csv. A traditional automated system would fail outright, trigger a generic error log, and halt your operations until an engineer logs in to fix the file path.
CoSet handles this with autonomous log analysis:
- Log Parsing: The Error Resolution Worker immediately catches the ingestion failure log.
- Contextual Matching: Rather than throwing its hands up, it analyzes the target directory. It recognizes that while the file string looks different,
Summit_Positions_07_07_2026matches the semantic intent and, crucially, respects the correct target date. - Self-Healing Execution: The agent safely matches the file, maps it to the workflow, and resolves the issue on the fly—ensuring your morning reconciliation completes without a hitch, while logging the adjustment for full transparency.
By pairing intentional architecture with agentic AI, CoSet transforms financial operations from a game of constant firefighting into a self-healing, streamlined workflow.