How SoftCo Compressed Roadmap Delivery from 4 Months to 30 Days
A two-day hands-on workshop turned informal, ad-hoc AI use into an AI-native way of working — compressing a four-month roadmap into 30 days and putting Claude Code in the hands of 30+ SoftCo engineers.
Executive summary
The problem
Engineers were using a variety of AI tools informally without explicit structure or guidelines. SoftCo’s core goals are accelerating development velocity by embedding AI meaningfully into existing SDLC workflows, and refactoring legacy code to be AI-native without disrupting ongoing projects.
The solution
We ran a focused two-day onsite workshop introducing Claude Code to SoftCo’s engineers using a real backlog item. Zartis then embedded NeXtGineers directly into SoftCo’s teams to sustain and scale adoption across two parallel workstreams: live product delivery, and a greenfield AI-enabled platform build.
The outcomes
- Development velocity increased 3x
- Roadmap delivery compressed from 4 months to 30 days (a 75% reduction)
- 76 user stories generated in ~2 hours (previously 3–6 weeks)
- Over 30 engineers actively using the Claude Code framework
- A verified production example: a feature estimated at 12–18 weeks shipped in 6–8 weeks
About the client
SoftCo is a software business operating in the fintech space, serving enterprise clients, reflecting years of continuous feature development for its customer base. As SoftCo looked to keep shipping new capability at pace without disrupting that installed base, it turned to AI-assisted development to accelerate delivery while modernising the underlying platform.
The problem
SoftCo’s core goals were to accelerate development velocity by embedding AI meaningfully into existing software development lifecycle (SDLC) workflows, and to refactor legacy code to be AI-native without disrupting ongoing projects.
Engineers were already using a variety of AI tools, but informally and without explicit structure or shared guidelines.
What Zartis brought
- Deep Claude Code expertise as a Preferred Services Partner in the Claude Partner Network.
- A structured, two-phase engagement methodology: hands-on workshop, then embedded delivery.
- Embedded senior engineers (NeXtGineers) working directly inside SoftCo’s teams, ensuring continuity of adoption beyond the initial workshop.
How We Worked Together
The Zartis Approach
Building on its existing relationship with SoftCo, Zartis kicked off with a hands-on workshop in late 2025: a focused, practical introduction to Claude Code for SoftCo’s engineers, working on a real backlog item.
The engagement was structured in two phases:
1. A focused two-day onsite workshop, where SoftCo's engineers worked hands-on with Claude Code on a real backlog item alongside cross-functional teams.
2. Embedded delivery, where Zartis NeXtGineers integrated directly into SoftCo's teams post-workshop to accelerate and sustain Claude Code adoption across the organisation.
Today, SoftCo has a team of 6 Zartis NeXtGineers embedded across two parallel workstreams:
Embedded delivery in existing product teams. Several NeXtGineers are integrated directly into SoftCo's established product teams, delivering live roadmap features and modernising a large, long-lived codebase. Claude Code is used throughout — writing and refactoring code, accelerating integration work (for example, supplier-onboarding and document-capture services), and increasingly for analysing and documenting complex legacy systems, so hard-won domain knowledge is captured rather than lost. This is the workstream in which SoftCo verified the velocity gains described above.
Greenfield platform and new AI-enabled products. A second group is building a new shared platform and a set of AI-enabled products from the ground up, applying an AI-in-the-loop approach across the full lifecycle — from product thinking and refinement through to QA and engineering. Beyond using Claude Code to build, this team has developed and published reusable Claude Code plugins and shared context standards, so AI usage stays consistent and repeatable and improves as the codebase grows, rather than fragmenting across individuals.
What this enabled next
What began as a single workshop has grown into a multi-team engagement spanning delivery, platform build, and ways-of-working. Several of these workstreams are only a few weeks old, so the current focus is establishing strong foundations and AI-native standards ahead of the next wave of measurable results.
The combination of deep Claude Code expertise, structured engagement methodology, and embedded senior engineers working directly in client teams is what transformed Claude from an interesting tool into a reliable part of SoftCo’s development workflow.
the results
Outcomes & Measurable Impact
At a glance
Impact during the workshop
The AI hackathon produced production-ready artifacts from a real backlog item – not demos.
These were outputs the team could continue building from immediately:
In the idea-to-prototype cycle
In ~2 hours (previously 3–6 weeks)
Reduced from 2-3 weeks to ~1 hour
Reduced from 1–2 weeks to ~1 hour
Compressed from 4 months to 30 days (75% reduction)
Reduced from 1–2 weeks to ~1 hour
Actively using the Claude Code framework
Shifted team perception of AI
From external tool to catalytic team member
Enabled true cross-functional collaboration
Across Product, Engineering, and QA simultaneously
Organic adoption beyond engineering teams
With non-technical employees independently building internal tools using Claude Code
Impact after the workshop
In practice, this means new capabilities and updates now reach SoftCo’s enterprise clients in weeks rather than months.
SoftCo has verified this shift with a real production example: a major roadmap feature had been estimated at 12–18 weeks. Once the team was fully enabled with Claude Code — following the late-2025 workshop — the feature shipped in 6–8 weeks instead, a 50–75% reduction in delivery time for a complex, production-ready feature.