Rebuilding the Amalgam Website in Four Days
We rebuilt the Amalgam website in four days to pressure-test our own delivery system.
The work combined bounded AI-assisted execution with human-owned architecture and QA decisions.
The result was a cleaner platform that ships faster and is easier to maintain.
Situation
Our previous site ran on WordPress and worked well early on.
As the company grew, performance tuning became fragile, design consistency slipped, and structural improvements started depending on plugin workarounds instead of system improvements.
We see this pattern often in client environments: the platform still works, but the underlying structure slows execution. Our own site had reached that point.
So we used the same approach we apply with clients: identify the real constraint, redesign the architecture, and ship the next system cleanly.
Diagnose the real problem
Before touching design or code, we reviewed the site as a system.
We looked at:
- page hierarchy
- navigation structure
- design consistency
- performance bottlenecks
- technical SEO signals
- long-term maintainability risk
Many redesigns skip this step. Changing visuals without fixing structure usually recreates the same problems later.
Move to a modern architecture
Once the constraints were clear, we rebuilt the site on Next.js.
This gave us much better control over:
- rendering strategy
- performance optimization
- technical SEO
- development flexibility
- long-term maintainability
WordPress remains great for many publishing workflows, but for tighter control of performance, architecture, and delivery speed, frameworks like Next.js provide a stronger base.
Build the design system first
Instead of designing page by page, we built the underlying system first.
That included defining:
- typography hierarchy
- layout grid and spacing rules
- component patterns
- responsive behavior rules
- color tokens and style rules
Once those foundations were set, building the rest of the site became much faster and more consistent. Speed without a system creates disorder; a design system preserves coherence.
Use AI models in clearly defined roles
A major part of the speed came from workflow structure: we used multiple models in bounded roles instead of one model trying to do everything.
Role examples:
- structural analysis
- UX critique
- component generation
- refactoring
- QA review
- documentation support
Different models are better at different tasks. Intentional orchestration accelerated delivery; unbounded overlap would have created inconsistency.
Keep humans responsible for judgment
AI accelerated execution, but it did not replace engineering discipline.
Humans remained responsible for:
- architecture decisions
- design judgment
- system consistency
- code review
- merge approval
- production readiness
Models accelerated options and iteration; final decisions stayed human.
Run everything inside controlled boundaries
Containment was non-negotiable. Models did not get unrestricted control over the codebase.
Work was scoped by stage, reviewed before merge, and validated for performance and structure before release.
This let us move quickly without hidden risk. AI acted as an execution layer inside a structured process, not an autonomous builder.
What changed
Speed
The rebuild, from architecture redesign to launch, happened in four days through parallel execution and faster iteration loops, not by skipping engineering discipline.
Performance
Moving to modern architecture removed core constraints: pages became faster, easier to optimize, and technical SEO became more predictable.
Consistency
The design system made the site more coherent: layout, spacing, and typography now follow shared rules instead of drifting page by page.
Maintainability
Internally, the system is now easier to extend and reason about, so future changes ship faster with less debt accumulation.
How this connects to our work
The sequence in this rebuild mirrors how we approach client pressure: understand the situation, isolate the real constraint, redesign the system around that constraint, and execute quickly inside clear review boundaries.
This same thinking now appears in a framework we recently launched on the site called Your Next Move, which helps builders understand where they are and what actually matters next.
Similar situations
We see this kind of challenge frequently with teams modernizing legacy platforms under delivery pressure.
- running WordPress sites that have become hard to optimize
- modernizing older CMS-driven platforms
- rebuilding marketing or product sites for performance
- exploring how AI can accelerate development safely
- trying to move faster without sacrificing engineering discipline