All case studies

From Excel Bug Tracking to Full QE Practice: Process Transformation at Capitol AI’s Enterprise Intelligence Platform

Capitol AI converts expert judgment and proprietary data into trusted, auditable outputs for enterprise data teams, financial services firms, and government bodies — all without vendor lock-in or data leakage. We didn’t just test the product. We changed how quality engineering works there.

Full-time QE Enterprise AI Jira Migration Multi-client Bi-weekly Releases
From Excel Bug Tracking to Full QE Practice: Process Transformation at Capitol AI’s Enterprise Intelligence Platform
About Capitol AI

Enterprise-grade agentic intelligence — built for regulated, high-stakes environments.

Capitol AI is an enterprise-grade agentic intelligence platform designed for organisations where accuracy, auditability, and sovereignty are non-negotiable. It converts expert judgment and proprietary data into trusted, customisable outputs — enabling enterprise data teams, financial services firms, professional services organisations, and government bodies to deploy AI workflows without vendor lock-in or data leakage risk. The platform is model-agnostic, SOC 2 compliant, and supports fully customisable output artifacts. It operates across multiple enterprise clients with independent environments per tenant — meaning that quality validation must account for both platform-level behaviour and client-specific configuration simultaneously. For a platform operating in regulated industries where a single incorrect output can carry significant consequences, quality engineering isn't a checkbox. It's a trust mechanism. The QE team's work at Capitol AI directly supports the reliability that enterprise clients depend on.

We didn't just test the product — we changed how quality engineering works at Capitol AI. Every structural improvement was initiated by the QE team.
Project Profile

At a glance.

Client
Onboarded
September 2024
Defect Tracking
Jira (migrated from Excel)
Release Cadence
Bi-weekly · 46 cycles
Industry
Enterprise AI / SaaS
Project Type
Full-cycle Manual QE (Full-time)
Engagement Model
Full-time (60 hrs / week — 40 + 20)
Environments
Dev, Staging, Production (multiple client tenants)
Release Cadence
Bi-weekly
Automation Partner
QA Wolf (automation identification and collaboration)
Test Coverage
13+ product modules, 251+ structured test cases
Engagement Journey

Built, structured, scaled.

The engagement moved through three distinct phases — an initial setup period characterised by high defect volume and zero process infrastructure, a process-build phase where every structural improvement was initiated by the QE team, and a scaled testing phase delivering sustained high-volume execution across bi-weekly releases and multiple client environments.

Period
Phase
Defects Raised
Key Milestones
Q1 2024
Setup
~120 defects
Informal Excel bug tracking; no release visibility; no test cases; establishing baseline understanding of the platform
Q2 2024
Process Build
245 defects · 73% closure
Jira migration complete; 251+ test cases created; Confluence docs; release communication windows established; dev–QA retest workflow introduced
Q3 2024
Scale
174 defects · 49% same-quarter closure
4,373 test executions; 46 release cycles supported; QA Wolf collaboration initiated; multi-client environment coverage (EY, Plexal)
The Challenge

Starting from zero on a live enterprise platform.

When the QE team joined Capitol AI, there was no structured quality engineering process in place. Bug tracking was done informally in Excel. Releases happened without any advance notice to QA. There were no test cases, no feature documentation, no recordings — and the platform was shipping features rapidly across multiple enterprise client environments.

01

No QA Process or Structure

Informal Excel-based bug tracking with no consistent format, no defect severity classification, and no process for how issues moved from discovery to resolution. The team was starting from zero on a live, enterprise-client-facing platform.

02

No Release Visibility

Releases went live with no advance notice to QA. The team had no ability to plan test cycles around release windows, meaning defects were often discovered post-release rather than pre-release. This needed to change immediately.

03

No Test Documentation or Feature Specs

No test cases, no feature matrix, no test plans. Developers were not creating feature documentation or recordings, leaving QA to reverse-engineer expected behaviour from the live product. Every test case had to be authored from scratch.

04

Rapidly Evolving AI Platform

Capitol AI ships new features and deprecates old ones frequently. Keeping test coverage aligned with a fast-moving enterprise AI product — while simultaneously building the process infrastructure — required continuous documentation discipline and close developer communication.

What We Did

Every structural improvement was initiated by QE.

The team didn't just test the product — they built the quality engineering foundation that Capitol AI now operates on. From defect workflow design to documentation culture, every structural change was driven and delivered by the QE team.

01

Built Test Documentation from Scratch

251+ structured test cases across 13+ product modules, built from ground-level product understanding. A full feature map was created first to establish coverage priority, then test cases authored to cover normal, edge-case, and multi-client scenarios across the platform.

02

Migrated Defect Tracking from Excel to Jira

The team designed and implemented a structured Jira workflow for defect tracking — replacing the informal Excel-based system entirely. Defects now have consistent severity ratings, reproduction steps, environment context, and a traceable history from discovery to closure.

03

Established Release Communication Process

Worked with the Capitol AI engineering team to define release windows and processes for QA advance notice. QA now receives release plans in advance — enabling structured pre-release test cycles rather than reactive post-release defect discovery.

04

Introduced Feature Documentation Practice

Drove the adoption of developer-created feature recordings and test plans, using Confluence as the documentation hub. This shifted the burden of requirements interpretation from QA inference to documented developer intent — reducing ambiguity and improving test coverage quality.

05

Structured Dev–QA Retest Workflow

Defined a clear, consistent process for how developers pick up resolved defect tickets, implement fixes, and return them to QA for retest. This eliminated the informal back-and-forth that had previously created delays and duplicate effort.

06

Sustained High-Volume Multi-Client Testing

4,373 test executions in Q3 alone across 46 release cycles, covering multiple client environments including EY and Plexal. QA Wolf automation collaboration initiated to identify automatable test cases and begin building an automation layer on top of the manual QE foundation.

Key Metrics

The numbers.

650+

Total Defects Raised

4,373

Q3 Test Executions

251+

Structured Test Cases

46

Release Cycles Supported

Results & Impact

What changed.

650+

Total Defects Raised

Across the full engagement — spanning Q1 baseline discovery through Q2 process build and Q3 scale phases.

4,373

Test Executions in Q3

Delivered in a single quarter across 46 bi-weekly release cycles and multiple enterprise client environments.

251+

Test Cases Created

Built from scratch across 13+ product modules — from feature mapping through edge-case authoring and multi-client scenario coverage.

46

Release Cycles Supported

Bi-weekly cadence across Q3, each with a defined QA plan, structured pre-release cycle, and documented defect history.

Q1 → Q3

From Zero Structure to Full QE Practice

Excel bug tracker to Jira; no release notice to defined release windows; no docs to Confluence hub — every structural change initiated by QE.

Jira

Migration Owned Entirely by QE

The defect tracking migration — from design to implementation to adoption — was driven and delivered by the QE team, not the engineering organisation.

Tools & Technology

The stack.

Defect Tracking
Jira — migrated from informal Excel; full defect lifecycle management
Documentation
Confluence — feature documentation hub; test plans; developer recording repository
Automation Partner
QA Wolf — automation identification and collaboration
Environments
Dev, Staging, Production (multiple client tenants — EY, Plexal, and others)
Release Cadence
Bi-weekly (46 cycles in Q3)
Test Coverage
13+ modules; normal, edge-case, and multi-client scenarios
Test Types
Functional, regression, exploratory, sanity, retesting, cross-client validation
The Takeaway

The Capitol AI engagement is a story of QE-led process transformation. The team didn't just test the product — they changed how quality engineering works at Capitol. From driving the migration to Jira, to establishing release communication norms, to introducing documentation culture, to structuring the dev–QA retest workflow — every structural improvement was initiated and delivered by the QE team. The result is a quality engineering practice that can sustain 4,373 executions per quarter, support 46 release cycles, and continue to evolve as the platform grows.