AI adoption built for measurement, not belief.
Syscon Inc. builds FIT, a field-tracking platform integrated with Sage for payroll and expense processing. As the product and its codebase evolved, the engineering team saw an opportunity to move beyond individual AI experimentation and establish a structured approach to AI-assisted development.
Together with Santex, Syscon put Anthropic’s Claude Code and Claude Cowork into production across its daily engineering workflows, pairing AI-assisted development with continuous measurement from the start.
Growth outpaced the team's ability to prove its own progress.
FIT had grown across three separate codebases — API, mobile, and ops (the management console) — while evolving toward a multi-tenant platform.
That complexity showed up directly in the development workflow. Pull requests were large and slow to review, holding finished work in the queue. Test coverage on the core mobile app was limited. Deployments and architectural decisions left no automated record, so the team had no reliable baseline for what was changing or how fast.
Built into the workflow, proven by continuous measurement.
Santex worked with Syscon to embed AI into the engineering workflow through a phased rollout, moving from assisted development toward increasingly autonomous execution. The rollout began with a small group of developers and expanded into the team’s standard way of working.
Claude Code was configured with custom plugins and subagents built around AI Engineering principles, enabling it to turn well-defined tickets into review-ready code while engineers remained in control of validation and delivery.
At the same time, four always-on measurement suites were introduced to continuously evaluate the impact on delivery, quality, releases, and engineering decisions.
“Building our own plugin and subagents around AI engineering principles turned Claude Code from a helper into a real force multiplier for the team.
Mariano B.
Lead AI Engineer
@ Santex
From individual AI adoption to a measurable engineering system.
Six months after Claude Code and Claude Cowork entered Syscon's daily engineering workflow, AI-assisted development is running in production — and the results are measured, not estimated.
Pull requests that were once large and slow to review now arrive review-ready and tightly scoped, cutting review time by 96% as measured by CHRONOS from commit to merge. Over the same six months, the team increased delivery throughput 3.5×. ARGOS tracks test and deployment quality as that velocity rose, SESHAT gives every release an automatic record of what shipped, and METIS holds 23 architecture decisions under continuous automated audit.
The outcome is not a pilot that produced a report. It is a production engineering workflow where AI-assisted delivery and continuous measurement run together, every day.
AI adoption built for measurement, not belief.
Syscon Inc. builds FIT, a field-tracking platform integrated with Sage for payroll and expense processing. As the product and its codebase evolved, the engineering team saw an opportunity to move beyond individual AI experimentation and establish a structured approach to AI-assisted development.
Together with Santex, Syscon put Anthropic’s Claude Code and Claude Cowork into production across its daily engineering workflows, pairing AI-assisted development with continuous measurement from the start.
Growth outpaced the team's ability to prove its own progress.
FIT had grown across three separate codebases — API, mobile, and ops (the management console) — while evolving toward a multi-tenant platform.
That complexity showed up directly in the development workflow. Pull requests were large and slow to review, holding finished work in the queue. Test coverage on the core mobile app was limited. Deployments and architectural decisions left no automated record, so the team had no reliable baseline for what was changing or how fast.
Built into the workflow, proven by continuous measurement.
Santex worked with Syscon to embed AI into the engineering workflow through a phased rollout, moving from assisted development toward increasingly autonomous execution. The rollout began with a small group of developers and expanded into the team’s standard way of working.
Claude Code was configured with custom plugins and subagents built around AI Engineering principles, enabling it to turn well-defined tickets into review-ready code while engineers remained in control of validation and delivery.
At the same time, four always-on measurement suites were introduced to continuously evaluate the impact on delivery, quality, releases, and engineering decisions.
“Building our own plugin and subagents around AI engineering principles turned Claude Code from a helper into a real force multiplier for the team.
Mariano B.
Lead AI Engineer
@ Santex
From individual AI adoption to a measurable engineering system.
Six months after Claude Code and Claude Cowork entered Syscon's daily engineering workflow, AI-assisted development is running in production — and the results are measured, not estimated.
Pull requests that were once large and slow to review now arrive review-ready and tightly scoped, cutting review time by 96% as measured by CHRONOS from commit to merge. Over the same six months, the team increased delivery throughput 3.5×. ARGOS tracks test and deployment quality as that velocity rose, SESHAT gives every release an automatic record of what shipped, and METIS holds 23 architecture decisions under continuous automated audit.
The outcome is not a pilot that produced a report. It is a production engineering workflow where AI-assisted delivery and continuous measurement run together, every day.
