00 A New Partner in an Unfamiliar Domain
When Tracy Garon Hojka joined Khojant LLC as Vice President in January 2026, she brought something most software companies would consider extraordinary: 30 years of financial data management experience, two decades in the Registered Investment Advisory sector, $450M+ in assets under management at Rockefeller Capital, and credentials including FINRA SIE and CIPM.
What she did not bring was any background in insurance.
The insurance industry operates differently from the investment world Tracy had spent her career in. The terminology is different. The data structures are different. The carrier relationships, the billing cycles, the commission structures — all different. And The Marathon System, built and evolved over more than 40 years by Jeff, carries an enormous amount of institutional architecture that lives nowhere except in the software itself and in the head of the person who built it.
Tracy's first task wasn't to build anything. It was to learn everything — and then prove that what she'd learned was correct by testing it to destruction.
The Onboarding Hurdles
Before Tracy could contribute meaningfully as a tester and data analyst, she had to clear a series of real, concrete learning hurdles — each one requiring time, patience, and deep curiosity:
Insurance Platform Fundamentals
Learning how independent insurance agencies actually operate — Agency Bill vs. Direct Bill, policy lifecycles, premium financing, and how carriers relate to agents.
Carrier Portal Reports
Each insurance carrier has its own portal with its own export formats, quirks, and naming conventions. Tracy had to learn how to pull the right data from each one reliably.
The Matching System
Understanding the 8-level reconciliation engine Jeff built — how it ingests data, how the matching tiers work, and what each exception type actually means operationally.
Error Communication
Learning how to describe software bugs in a way a developer can act on — isolating the condition, documenting the steps, and separating user error from system error.
Regression Testing
Understanding how to re-validate existing functionality after every new change — and how to design test scenarios that expose edge cases before real customers do.
40 Years of Architecture
The Marathon System has decades of internal logic, data schemas, and workflow design. Getting enough of it into her head to test it meaningfully was its own project.
What made Tracy's transition work wasn't that the learning curve was easy — it wasn't. It's that her 30 years of financial data instincts transferred directly. She knew what clean data looked like. She knew when a number didn't make sense. And she knew how to document discrepancies with the rigor of someone who's spent two decades producing audit-ready books.
Those skills turned out to be exactly what insurance reconciliation testing needed.
// Project 01
01 The Insurance Matching & Reconciliation Engine
⚠ The Challenge
Insurance statement reconciliation historically consumed hours of manual validation per cycle. Carriers deliver data in wildly different formats — complex multi-page PDFs, proprietary spreadsheets, and inconsistent naming conventions. Missing transactions, dropped policy prefixes, and format mismatches caused immediate revenue leakage, particularly in Agency Bill configurations where precise commission tracking is non-negotiable.
✓ The Solution
Jeff and Tracy engineered an intelligent parsing and reconciliation engine built directly into The Marathon System. The system normalizes raw inbound data from any carrier into a unified financial schema, then runs it through a multi-level matching algorithm that resolves as much as possible automatically and escalates only what requires human judgment.
The Ingestion Pipeline
The ingestion layer uses custom text extraction algorithms to strip formatting artifacts from carrier PDFs. Target data is then processed through rigid regex pattern matrices to isolate policy numbers, premium amounts, effective dates, and commission figures regardless of how the carrier formatted the original document.
Normalized output maps into six standardized structural categories that preserve the operational history of every transaction:
Production · ProductionCalc · Statements · Sweeps · Fees · Statement Payments
The 8-Level Matching Engine
At the heart of the system is an algorithmic logic matrix that guides reconciliation through eight tiers of matching confidence — from trivial exact matches down to items that require manual intervention:
Workflow tooling complements the engine: native inline corrections allow real-time changes to wrong-company entries, account flags append tracking data directly to source files, and bulk action tools clear high-volume exceptions rapidly without navigating record by record.
Tracy's Role: Testing What 40 Years Built
This is where Tracy's onboarding investment paid off. Testing an 8-level reconciliation engine requires someone who can construct scenarios that break it — not randomly, but intelligently. Tracy brought her financial reconciliation instincts from the investment world and applied them to insurance data, designing test cases that surfaced real edge cases from real carrier behavior.
One early example: AI analysis isolated and patched a complex bug involving inconsistent carrier policy prefix formatting. The system was failing to match valid policies because different user entry behaviors produced different prefix formats for the same underlying policy number. Tracy identified the pattern. Jeff patched it with a localized truncation fix that normalizes the identifier before comparison. The system now handles it automatically.
Results
Reconciliation timelines dropped from hours to minutes. The system achieved an immediate 90% exact-match rate across complex carrier premium and commission tables on initial deployment. It is currently active across three customer environments, with a fourth in onboarding.
// Project 02
02 Corporate Brand Modernization & Infrastructure
The second initiative addressed something that had been deferred for years: the public digital presence of Khojant LLC and The Marathon System didn't reflect the depth of what the company had become.
AI-augmented pipelines structured decades of career history, technical achievements, and high-net-worth data expertise into polished, SEO-optimized marketing copy across three interconnected digital properties:
Khojant Corporate Hub — khojant.com
Rebuilt as a primary business consultancy and software development hub. Features formal case studies, cross-industry financial data capabilities, and executive profiles for both Jeff and Tracy that balance deep professional credentials with authentic personal background.
The Marathon System Product Portal — mscflexone.com
Engineered as a customer acquisition engine with full technical SEO, explicit Bing optimization, integrated performance analytics (GA4), automated lead-capture chat, and cryptographic bot protections. Production web environments are built on custom Ubuntu Linux servers, with AI assisting in mapping legacy Xenix administration methodologies onto modern Linux architectures.
LinkedIn Executive Profiles
Personalized professional profiles for both Jeff and Tracy, positioning Khojant's capabilities across insurance technology, financial data management, AI-augmented workflows, and software QA — building a professional footprint that matches the company's actual depth.
// Project 03
03 Automated Corporate Help System
The final phase of the winter work addressed a friction point that had existed for as long as The Marathon System had been in use: onboarding new agencies required significant hands-on time from Jeff, because the documentation simply didn't exist in a form users could access themselves.
The new help system is built directly into the Marathon application framework — not a separate knowledge base or external wiki, but contextual documentation that lives inside the same environment users are already working in.
Comprehensive Coverage
Every operational data field, complex transaction form, system layout, and financial report is documented. New carriers, new reconciliation scenarios, new edge cases — all covered before agencies encounter them in production.
Interactive Previews
Dynamic sandbox-style windows let users review live structural examples of generation forms, system letters, and export configurations before processing production data — eliminating the most common class of onboarding errors.
Universal Export Tools
Integrated universal data structures across all major system data grids, enabling instant extraction to .csv, .html, or .json formats. Agencies can now pull their own data into any reporting tool they prefer without needing custom exports from Khojant.
04 What This Winter Proved
The Winter Transformation wasn't just about three software projects. It was a proof of concept for what Khojant can do when deep insurance engineering experience pairs with deep financial data expertise — and when both sides are willing to do the hard work of learning each other's domain.
Tracy's onboarding journey is worth noting plainly: she came in with extraordinary credentials and zero insurance experience, absorbed 40 years of system architecture, learned to navigate carrier portals, mastered the matching engine, and became a rigorous regression tester — all while the system she was testing was simultaneously being extended. That is not a small thing.
The reconciliation system now processes in minutes what used to take hours. The digital infrastructure now reflects what the company actually is. And the help system means new agencies can onboard themselves.
Khojant enters the next phase of its development as a stronger company than it was last fall — and that is entirely by design.
"Expertise transfers. You just have to be willing to do the learning."
— Tracy Garon Hojka, Vice President · Khojant LLC
// about the authors
Jeff Hojka
35+ years building mission-critical software for independent insurance agencies. Founder of Khojant LLC and architect of The Marathon System — a cloud-hosted office suite running since 1982. Specializes in Clarion development, cloud infrastructure (AWS, HyperV), and AI-augmented software development.
Tracy Garon Hojka
30+ years of financial data management across RIA firms and family offices. Former Manager of Operations at Rockefeller Capital Management — $450M+ AUM, 1,800+ accounts. FINRA SIE certified. CIPM credentialed. Now leads reconciliation engineering, QA, and data analysis at Khojant LLC.