LATAM travel and expense platformTravel & Expense

Blocked backlog to production delivery and a technical baseline in 30 days

Agentify joined a regional travel and expense platform to do two things at once: unblock stalled development work and build a grounded plan for modernizing the product -- without stopping operations.

Published as

Names and identifying details have been removed. All metrics come directly from the first-month report -- nothing softened or rounded.

Case window

First 30 days

Most engagements start with an audit phase and deliver later. We skipped that split. In the first month, the team got into the codebase, learned how the business operates, closed priority tickets, and produced a scored technical baseline that could guide modernization decisions with evidence rather than gut feel.

The situation before Agentify

The organization had to keep shipping features on a platform built from legacy layers, fragmented knowledge, and years of accumulated technical debt.

Pending tickets cut across the database, business logic, API layer, and WebForms views.
Eighty-five percent of the codebase was still on .NET Framework 4.7.2, with data access relying on 421+ stored procedures.
Unit test coverage was 0%, and there was no original technical documentation to reduce regression risk.
Prioritization involved multiple stakeholders across business and operations, and the context needed to make those decisions was spread across people and tools.

The first 30 days

We started delivering value during onboarding, not after it.

Week 1

Business and technical kick-off

We started by understanding product vision, commercial priorities, the current architecture, and the status of migration efforts already in motion.

  • Alignment with business stakeholders on priorities and bottlenecks.
  • Technical immersion into .NET WebForms, SQL Server, stored procedures, and the Entity Framework migration path.

Week 2

Environment, repository, and functional context

We set up the local environment, mapped structure and dependencies, and automated the extraction of operating documentation so the team could ramp up faster.

  • Full environment setup and structural read of the repository.
  • Automated Zendesk scraping and creation of a functional-context skill to speed future AI-assisted cycles.

Week 3

Stakeholders and critical tickets

We ran stakeholder interviews in parallel with backlog work -- delivery started before the picture was complete.

  • Interviews with business and operations stakeholders to map real workflows.
  • Start of ticket 547 and a tool evaluation report for support operations.

Week 4

Intensive delivery + modernization baseline

With enough context in hand, we accelerated on priority tickets and closed out a technical diagnostic that would shape the next decisions.

  • Tickets 547, 553, and 493 merged to production.
  • Progress on the SFTP 558 integration, plus an eight-dimension technical scorecard that set the starting baseline.

What we found once inside

Rather than talking about technical debt in the abstract, we measured it: complexity, dependencies, integrations, and maintainability -- scored and documented.

421+

stored procedures referenced

155+

tables mapped

51+

Azure Functions analyzed

3.5/10

maintainability

Findings that shaped the priorities

  • The layered architecture showed some consistency, but maintenance risk was critical because there were no tests at all.
  • God classes like Rendicion.cs (11,780 lines) concentrated sensitive business logic in a single file.
  • There were legacy and conflicting dependencies, including DotNetOpenAuth with no maintenance since 2016.
  • The Entity Framework migration had started, but it covered only a small fraction of what needed to move.
  • Critical technical knowledge lived in people's heads, chat threads, and scattered tools -- there was no reliable documentation system.

What shipped in the first month

The first month mixed discovery, delivery, and early modernization work. There was no dead phase where we only studied the system.

Priority backlog unblocked

The team fixed tickets that touched operational logic and user experience across multiple layers of the product.

  • Ticket 547: fix to participant icon display logic.
  • Ticket 553: payment blocking for fixed funds across entities.
  • Three PRs merged to production during the first month.

Incremental data-access modernization

While working through backlog, the team also advanced one structural piece of the migration and established a pattern others could reuse.

  • Full migration of PrecioKmCategoria to Entity Framework.
  • Removal of 4 legacy stored procedures.
  • Rewrites in DataAccess and API to support the new data-access path.

New integration capability

We opened a new front to automate accounting exchange with a third-party system, without waiting for the broader modernization to finish.

  • TecnoAccion SFTP integration for accounting files covering advances and expense reports.
  • Progressive refactor of the CSV generation flow.
  • Coverage from business logic down to WebJobs and technical configuration.

A diagnostic that informs real decisions

The system readout went beyond scattered notes -- it turned into scored artifacts the team could use to prioritize.

  • 19 analysis documents produced.
  • Eight-dimension technical scorecard with a 5.8/10 global baseline.
  • A 30, 90, and 365 day roadmap for incremental modernization.

Visible impact in the first month

The first month produced results across three areas: execution speed, technical health, and the organization's ability to make better decisions going forward.

Execution impact

The team was productive from the start, shipping real output while still learning the system.

24

commits during the period

45.3%

of total repo output

4

tasks addressed

  • Agentify contributed 45.3% of the period's commits while operating as 2 developers inside a team with 6 active contributors.
  • During the peak week, an Agentify developer was the single most active contributor across the whole team.
  • Four workstreams ran in parallel: 3 were closed and 1 was still in progress at the time of the report.

Technical impact

Beyond shipping fixes, structural pieces moved in ways that reduce friction going forward.

4

stored procedures removed

6

layers covered by delivery

63.1%

of code touched in period

  • Delivery touched the full stack -- from database to WebJob and WebApp -- rather than staying inside a single layer.
  • The Entity Framework migration went from a vague intention to a working pattern that the team can repeat.
  • The technical baseline replaced gut-feel assessments with a prioritized map of risk, dependencies, and maintainability.

Organizational impact

The work also reduced internal friction and left better conditions for the next cycle.

19

analysis documents

16+

integrations mapped

1

modernization roadmap

  • Stakeholder interviews across business and operations reduced context loss between teams.
  • The functional documentation we extracted and structured became reusable context for future development cycles, including AI-assisted work.
  • The phased roadmap gave the organization a clear way to decide what to modernize first and why.

Why this case matters

The difference here was straightforward: technical judgment, speed, and knowledge transfer from day one -- not a promise of some future transformation.

01

We work inside the real system

We entered the product as it already existed -- stakeholders, constraints, accumulated debt and all. No sandbox. No fictional backlog.

02

We diagnose while we ship

While building the technical baseline, we were also merging tickets to production. That means stakeholders see value in weeks, not months.

03

Modernization starts before the big migration

Instead of promising a total rewrite, we moved one entity to Entity Framework, removed legacy stored procedures, and left a reusable pattern for the next phase.

04

We leave the team stronger than we found it

After interviews, context tooling, and 19 analysis documents, the knowledge that used to live in people's heads and chat threads was documented and accessible.