How Fundxyz Revolutionized Investment Operations with Swfte

Case study on how Fundxyz automated complex human operation processes with Swfte's custom AI agents and workflows

English

TL;DR

When Fundxyz, a leading global investment firm, approached us, they were drowning in manual processes that were consuming over 60 hours of their team's time weekly. Their investment analysts were spending precious hours on repetitive tasks rather than on high-value strategic thinking and decision-making.

In this case study, I'll walk you through how Fundxyz used Swfte to transform their operations, saving countless hours and dramatically improving their investment processes.

The Challenge

Before diving into implementation, the Fundxyz team identified their most pressing operational challenges:

  • Manual Data Processing

    Analysts were spending 20+ hours per week manually collecting, cleaning, and processing data from various financial sources.


  • Inconsistent Workflows

    Each analyst had their own way of working, leading to inconsistent processes and making handoffs between team members inefficient.


  • Documentation Bottlenecks

    Creating investment memos and compliance documentation was taking up to 15 hours per investment opportunity.


The Solution

What really made the difference for Fundxyz wasn't just implementing a single solution, but building a custom ecosystem of AI agents and workflows with Swfte.

Building Custom Agents

Fundxyz created three specialized agents with distinct roles:

model DataAgent {
  id          String

  capabilities [
    "financial_data_extraction",
    "data_cleaning",
    "pattern_recognition"
  ]

  integration  [
    "bloomberg_terminal",
    "refinitiv_eikon",
    "internal_databases"
  ]

  context      String
}
scheme
UTF-8

Let's break down each agent they created:

  • DataExtractionAgent

    This was their first agent, designed to automatically scan, extract, and normalize data from financial reports, news, and market data. Swfte's no-code interface made it easy to connect to their existing financial data sources.

    Using Swfte's model selection feature, they chose a model specifically optimized for financial document understanding. The team simply uploaded examples of the data formats they typically work with, and within hours, they had a working agent that could recognize patterns across diverse financial documents.


  • AnalysisAgent

    The second agent was built to process the normalized data and generate initial investment analyses. It was trained on Fundxyz's historical investment decisions to understand their specific evaluation criteria.

    What impressed the team most was how easily they could customize this agent without writing code. They used Swfte's playground to test various prompts and scenarios, quickly iterating until the agent produced the quality of analysis they needed.


  • DocumentationAgent

    This agent automated the creation of investment memos, compliance documentation, and client-facing materials. It pulled data from the other agents and formatted it according to Fundxyz's templates.

    The compliance team especially appreciated how this agent maintained an audit trail of every piece of information it included, making regulatory reviews much smoother.


Creating Integrated Workflows

With the agents built, Fundxyz used Swfte's workflow builder to create end-to-end processes:

export const investmentWorkflow = createWorkflow({
  name: 'Investment Evaluation Process',
  description: 'End-to-end workflow for evaluating potential investments',
  agents: [DataExtractionAgent, AnalysisAgent, DocumentationAgent],
  triggers: {
    manual: true,
    scheduled: '0 6 * * 1-5', // Weekdays at 6 AM
    api: true,
  },
  steps: [
    {
      name: 'Extract Data',
      agent: 'DataExtractionAgent',
      action: 'processFinancialDocuments',
      inputMapping: {
        documents: '{{trigger.documents}}',
        dataTypes: ['financials', 'market_trends', 'competition'],
      },
    },
    // Additional steps defined here
  ],
});
TypeScript
UTF-8

Their workflow design centered around three core processes:

  1. Data Collection and Normalization

    The DataExtractionAgent would activate whenever new investment opportunities were uploaded to their system. It would extract all relevant financial data, normalize it, and store it in their database.

    What made this particularly powerful was how the agent could adapt to different document formats without requiring new training. Whether dealing with annual reports, pitch decks, or market analyses, the agent could identify and extract the relevant information.


  2. Analysis and Scoring

    Once data was normalized, the AnalysisAgent would automatically generate a preliminary analysis and investment score. This wasn't meant to replace human judgment but to give analysts a head start and ensure consistent evaluation criteria.

    The team loved how they could trace the agent's reasoning. Swfte's observability features let them see exactly which factors influenced each score, making it easy to fine-tune the agent's parameters when needed.


  3. Documentation Generation

    The final step in their workflow automated the creation of investment memos and compliance documentation. The DocumentationAgent would pull data from previous steps, format it according to their templates, and flag any missing information.

    This step alone saved hours per investment opportunity and dramatically reduced errors in their documentation.


Implementation Journey

Fundxyz's journey to full automation wasn't overnight. Here's how they approached it:

Phase 1: Agent Creation

The first month was spent building and refining their agents. Using Swfte's no-code interface, their team was able to:

  • Select the right models for each agent's specific task
  • Provide example data for training and fine-tuning
  • Test agents in the playground environment
  • Connect agents to their existing data sources

What surprised them was how quickly they could iterate. With Swfte's instant feedback and testing tools, they could see results immediately and refine their agents accordingly.

Phase 2: Workflow Integration

With the agents built, they turned to creating integrated workflows. This phase focused on:

  • Designing the logical flow between agents
  • Setting up triggers for automation
  • Creating feedback loops for continuous improvement
  • Implementing approval steps where human oversight was needed

The team particularly appreciated Swfte's visual workflow builder, which made it easy to experiment with different process designs.

Phase 3: Scaling and Optimization

Once the basic workflows were in place, Fundxyz focused on optimization:

  • Adding more specialized agents for specific investment types
  • Creating dashboards to monitor workflow performance
  • Implementing A/B testing for different analysis approaches
  • Gradually reducing human intervention points as confidence increased

Throughout this phase, they relied heavily on Swfte's observability tools to identify bottlenecks and optimization opportunities.

Final Results

After six months of implementation and refinement, Fundxyz achieved remarkable results:

0
0
0
0

The numbers tell a compelling story:

  • 75% reduction in time spent on data processing
  • 88% faster generation of investment memos
  • 3x increase in the number of opportunities analysts could evaluate
  • 95% reduction in documentation errors

But the qualitative benefits were equally important:

  • Analysts reported higher job satisfaction, now focusing on strategic analysis instead of data wrangling
  • The compliance team gained better visibility into the investment decision process
  • The consistency of their evaluation process improved dramatically
  • New team members could be onboarded much faster with standardized workflows

As the CTO of Fundxyz put it: "Swfte hasn't just saved us time; it's fundamentally transformed how we approach investment analysis. Our team is making better decisions faster, with more confidence and less busywork."

Summary

Fundxyz's success with Swfte demonstrates the power of custom AI agents and workflows in the financial sector. By automating routine tasks and creating consistent processes, they've been able to focus their human talent on what matters most: making strategic investment decisions.

The key factors in their success included:

  • Using specialized agents for different parts of their process
  • Creating integrated workflows that connect these agents
  • Gradually implementing automation with appropriate human oversight
  • Continuously monitoring and optimizing their processes

For any investment firm looking to reduce operational overhead and improve analyst productivity, Swfte offers a powerful platform for building custom AI solutions without writing code.

Interested in learning how Swfte could transform your operations? Schedule a demo to see our platform in action.

Posted onsuccess-storieswith tags:
#case-study#finance#automation
0
0
0
0