Month-end close is one of the most critical and painful processes in finance.
Reconciling contract data, CRM records, and billing systems is slow, manual, and error-prone. For our finance team at SnapLogic, it meant spending more than 40 hours every month manually validating customer order forms against Salesforce data, with no single source of truth to rely on.
In the second episode of our Agentic Builders webinar series, I partnered with Senior Finance Manager, Nicole Houts, to show how we tackled this challenge by building a finance agentic workflow.
The results: a 90% reduction in manual review time, a 30% faster monthly close, and even 2% of annual revenue recovered.
The finance challenge
Nicole shared what our finance team was up against:
- Manual reconciliation between Salesforce opportunities and PDF contracts stored in Box
- 40+ hours a month wasted on tedious validation and error correction
- Fragmented systems with no single source of truth for critical financial data
- Error risk and revenue leakage from relying on manual checks
As Nicole put it during the session: “Anytime you have manual processes, they are error-prone. That can result in incorrect reporting and even revenue leakage — none of which are ideal for a finance team.”
The target outcome was clear: save time, improve accuracy, and free up the finance team to focus on more strategic, high-value work.
Building the Finance agentic workflow
To solve this, we designed a finance agent powered by SnapLogic’s agentic integration platform. Here’s how we approached it:
- Interface: We built the agent on Streamlit, providing a simple, chat-style UI for finance users to kick off reconciliations, review results, and iterate quickly.
- Architecture: SnapLogic’s driver-worker pattern allowed us to orchestrate LLMs with enterprise context while keeping humans in the loop when needed.
- Connected tools:
- Salesforce for opportunity and revenue data
- Box for customer order forms
- Excel outputs for reconciliation reports
The workflow starts by retrieving closed-won opportunities from Salesforce, pulling the associated order forms from Box, and reconciling key financial data. Discrepancies are flagged instantly, links to records are provided for quick validation, and results can be reviewed either in Excel or directly in the agent interface.
Demo highlights
During the live demo, we showed how quickly the agent could:
- Retrieve renewal opportunities from Salesforce and provide direct record links
- Cross-reference order forms stored in Box
- Highlight mismatches in financial details across systems
- Generate consolidated Excel reconciliation reports for final review
As Nicole said in the session: “This is already light years ahead of where we were before. Just being able to see opportunities with links and reconciled data in one place saves so much time.”
Business impact
The results of putting this workflow into production were significant:
- 90% reduction in manual review time
- 30% faster monthly close
- 2% of annual revenue recovered from discrepancies that would otherwise have been missed
Perhaps equally important, the finance team shifted from tedious data entry to higher-value analysis and decision-making. By automating reconciliation, we gave time back to the people closest to the business.
Lessons learned along the way
Two big insights emerged from this project:
- Digging into the details is essential. Successful agent design starts with a deep understanding of the business process and pain points. Working closely with Nicole and the finance team helped us uncover the granular steps where automation could deliver the most value.
- Don’t be afraid to challenge the status quo. As we built the workflow, we discovered inefficiencies in the process itself. Streamlining and rethinking steps not only improved the agent, it improved the overall operation.
And one surprise: the agent didn’t just save time. It uncovered missed revenue. That’s the kind of outcome that turns a good project into a transformative one.
Perspectives d'avenir
Building agentic workflows can feel intimidating. Where do you start? How do you design something that both business users and technical teams will trust? What this project proved is that SnapLogic makes it easy to take that first step. By connecting critical data sources like Salesforce and Box, you create a springboard for success. From there, workflows evolve: adding tools, iterating on prompts, and scaling into production.
This finance use case is just the beginning. We’re already working on additional agents across finance and accounting, and the opportunities are vast. The bigger message is this: you don’t need to boil the ocean to get value from AI. Start with the data you already have, connect it with SnapLogic, and let agentic integration do the heavy lifting.
If you’re curious to learn more about how to build your own agentic workflow and need some inspiration to get started, register for our next Agentic Builders webinar or catch the recordings.
