Why AI Kit

A side-by-side comparison of AI-assisted development with and without AI Kit — across real workflows that happen every day.

The 60-Second Case

Without AI Kit, your AI assistant starts every conversation from scratch. It doesn’t know your framework, conventions, team standards, or project structure. You re-explain context every time. Output is inconsistent. Code reviews catch AI-generated violations.

With AI Kit, the AI already knows everything. One command. Zero ongoing effort.


Time to set up:       npx @mikulgohil/ai-kit init (30 seconds)
Time to maintain:     npx @mikulgohil/ai-kit update (when stack changes)
Runtime cost:         Zero — no background processes, no API calls

Side-by-Side: Daily Workflows

1. Creating a New Component

	Without AI Kit	With AI Kit
What you type	”Create a ProductCard component"	"Create a ProductCard component” or `/new-component ProductCard`
What AI knows	Nothing about your project	Your framework, styling, CMS, conventions — and which skill applies
What you get	Generic React component. Wrong export style, wrong file structure, no Sitecore helpers, no TypeScript patterns	Component matching your exact project patterns. Correct exports, Tailwind classes, Sitecore field helpers, typed props
Tests	None unless you ask	Auto-generated: happy path, error states, edge cases
Documentation	None	Auto-generated `.docs.md` if component is complex
Follow-up work	15-30 min fixing patterns, adding types, creating tests	Ready for review

2. Fixing a Bug

	Without AI Kit	With AI Kit
What you type	”fix the checkout bug"	"fix the checkout bug”
What happens first	AI guesses which file, reads 10+ files, tries random fixes	AI asks: “Which file? What’s expected vs actual?”
The fix	May fix the wrong thing. No test. No docs update.	Targeted fix + regression test + change log update
Review cycles	2-3 rounds catching missing tests, wrong patterns	1 round focused on business logic
Same bug returns?	Likely — no regression test	No — test prevents it

3. Pre-PR Review

	Without AI Kit	With AI Kit
What you do	Push code, open PR, wait for review	Run `/pre-pr` before pushing
What reviewers catch	Console.logs, missing types, `any` usage, no alt text, wrong imports, missing tests	Already fixed before they see it
Review time	30-60 min per review cycle	10-15 min focused on logic
Review cycles	2-4 rounds	1-2 rounds
Developer frustration	”I keep getting the same feedback”	Feedback is about architecture, not lint

4. Implementing a Figma Design

	Without AI Kit	With AI Kit
Approach	Eyeball the design, hardcode values	`/figma-to-code` — structured token mapping
Spacing/colors	Hardcoded `p-6`, `#1a2b3c`	Maps to design tokens: `p-spacing-lg`, `text-primary`
Design review cycles	3-4 rounds (“wrong spacing”, “wrong color”, “not responsive”)	1-2 rounds
Maintainability	Breaks when design system updates	Token changes propagate automatically

5. Onboarding a New Developer

	Without AI Kit	With AI Kit
Day 1	Read the wiki (if it exists). Ask teammates about conventions.	Run `ai-kit init`. AI already knows everything.
First PR	5+ review comments about naming, structure, missing docs	AI enforced conventions. Review focuses on logic.
Time to productivity	1-2 weeks	2-3 days
Questions asked	”What’s the naming convention? Where do tests go? Which import style?”	AI answers these through `/understand` and `/prompt-help`

6. Sitecore Component Development

	Without AI Kit	With AI Kit
Field helpers	Developer forgets `<Text>`, uses `{fields.title.value}` — breaks Experience Editor	Rules enforce field helpers. AI uses `<Text>`, `<RichText>`, `<Image>` automatically
Debugging	Hours googling “Sitecore component not rendering”	`/sitecore-debug` — structured checklist finds the issue in minutes
Component registration	Forgot to register in componentFactory — blank page	AI checks registration as part of `/new-component`

Impact by Role

For Individual Developers

Metric	Without AI Kit	With AI Kit	Improvement
Time explaining context to AI	5-10 min per conversation	0 min (auto-loaded)	100% eliminated
PR review cycles	2-4 rounds	1-2 rounds	50-75% fewer
Bug recurrence	Common (no regression tests)	Rare (auto-tested)	Significant reduction
Component creation time	30-45 min (fix patterns after)	10-15 min (right first time)	60-70% faster
Documentation debt	Grows constantly	Stays current (auto-enforced)	Eliminated

For Tech Leads

Metric	Without AI Kit	With AI Kit	Improvement
Code review time	30-60 min per PR	10-15 min per PR	50-75% reduction
Review comments about conventions	60% of all comments	Near zero	Standards auto-enforced
Onboarding time	1-2 weeks	2-3 days	70-80% faster
Cross-project consistency	Each project different	Same standards everywhere	Fully standardized
Standards documentation	Wiki nobody reads	CLAUDE.md AI actually follows	Finally enforced

For the Organization

Metric	Without AI Kit	With AI Kit	Improvement
AI tool ROI	Low — inconsistent output	High — reliable, standards-compliant	Dramatically higher
Knowledge sharing	Tribal knowledge	Codified in rules + guides	Preserved
Client code quality	Varies by developer	Consistent across team	Uniform quality
Technical debt from AI	Grows (AI generates non-standard code)	Minimal (AI follows standards)	Prevented
Security mistakes	Caught in review (if at all)	Caught by `/security-check` before commit	Shifted left

Real Scenario: A Week Without vs With AI Kit

Monday — Without AI Kit


09:00  Start new feature. Ask AI to create 3 components.
09:30  AI generates components with wrong patterns. Start fixing.
10:30  Components match conventions. Start writing tests.
11:30  Tests done. Push PR.
13:00  Review feedback: "Missing alt text on images. Use named exports,
       not default. Add docs for the data table component.
       The ProductCard should use Sitecore field helpers."
14:00  Fix all review feedback. Re-push.
15:00  Second review: "The regression test for the bug fix is missing.
       Also, there's a console.log on line 45."
15:30  Fix, re-push.
16:00  Approved on third cycle.

Total: 7 hours. 3 review cycles. Multiple convention violations.

Monday — With AI Kit


09:00  Start new feature. Run /new-component for each.
09:45  AI generates components with correct patterns, tests, and docs.
       Sitecore field helpers included automatically.
10:00  Run /pre-pr. Catches a missing alt text and a console.log.
10:15  Fix both. Push PR.
11:00  Review feedback: "Could we use a discriminated union for the
       loading state?" (Actual architecture feedback, not conventions.)
11:30  Update the type. Re-push.
12:00  Approved on second cycle.

Total: 3 hours. 2 review cycles. Zero convention violations.

What AI Kit Does NOT Do

Being honest about limitations:

Claim	Reality
”AI Kit makes AI perfect”	No. AI still makes mistakes. But it makes fewer, more specific ones.
”You never need code review”	No. AI Kit catches conventions and patterns. Humans review logic and architecture.
”It works without developer effort”	Partially. The rules auto-enforce standards, but developers still need to write clear prompts for complex tasks.
”It replaces documentation”	No. It generates and maintains docs automatically, but developers should still write architectural decision records and design docs.
”100% consistent output”	~90%. Prompt engineering gets close, but AI models have natural variance. Edge cases still require human judgment.

Cost-Benefit Summary

	Cost	Benefit
Setup	30 seconds (`npx @mikulgohil/ai-kit init`)	AI understands your project forever
Maintenance	Occasional `ai-kit update`	Rules stay current with stack changes
Learning curve	Read getting-started.md (5 min)	48 skills auto-discovered — no command names to learn
Runtime overhead	Zero — static files only	No performance impact
Team adoption	Commit generated files to git	Every developer gets the same AI context

The question isn’t “is AI Kit worth it?” — it’s “why would you use AI tools without it?”

Get Started


npx @mikulgohil/ai-kit init

See Getting Started for the full walkthrough, or read AI That Improves Itself to understand the technology behind these improvements.