How I used Codex App to quickly analyze a securities issuer

Case context

The case began with a practical question: how to quickly assess a new issuer in Nicaragua’s securities market without losing source traceability. The goal was not to produce an investment recommendation, but to test how an agentic AI tool could accelerate an initial document review.

OpenAI Codex App

I used Codex App to help research public documents, organize evidence, cross-check sources and turn the information into a preliminary assessment that could be reviewed.

What the AI was asked to do

Analyze the issuer in depth, review public documentation, identify favorable signals, risks and due diligence questions, and produce an initial assessment for human review.

What the tool did

• Reviewed the issuer profile on the Nicaragua Stock Exchange website.
• Searched SIBOIF for regulatory information about the issuer.
• Reviewed the credit rating and qualitative signals in the PCR report.
• Consulted audited financial statements and recent public data.
• Calculated or summarized relevant indicators of growth, delinquency, coverage, liquidity, profitability and solvency.
• Separated positive signals, warnings, unresolved conditions and due diligence questions.

Analysis result

The generated result was an AI-assisted preliminary assessment: Ficohsa Nicaragua appeared to be a reasonable banking issuer to study, but a complete assessment depended on each series’ specific terms, particularly its rate, maturity, structure, secondary-market liquidity and final documentation.

• Banking issuer rated NI AA- with a stable outlook, according to PCR.
• Authorized program of up to US$50 million for commercial paper and bonds.
• General backing from the issuer, without specific collateral attached to the instrument.
• Asset and loan-portfolio growth observed in the public information reviewed.
• Controlled delinquency and coverage of past-due loans and loans in judicial collection as favorable signals.
• Profitability, efficiency, relative solvency and IT findings as areas requiring monitoring.
• A need to establish the rate, maturity, payment frequency, amortization, secondary-market liquidity and specific terms of each series.

What this use case teaches

• AI does not replace human judgment, but it accelerates the collection and cross-reading of public information.
• The agent helps turn long documents into signals, risks and actionable questions.
• Its greatest value comes when the result is structured for review, rather than accepted as a final conclusion.
• A useful case does not end with an answer; it ends with better due diligence questions.

Questions the tool helped formulate

• What is the interest rate or coupon for the specific series?
• What are the maturity and payment frequency, and is repayment amortizing or a bullet payment?
• Is there a repurchase or early-redemption option?
• What is the payment ranking, and how does the issuer’s general backing work?
• What volume will actually be placed, and what secondary-market trading history can be expected?
• Has the credit rating been updated since the audited 2025 financial statements?

Public sources