cd /news/artificial-intelligence/git-knows-who-ai-knows-what-nobody-k… · home topics artificial-intelligence article
[ARTICLE · art-19133] src=dev.to pub= topic=artificial-intelligence verified=true sentiment=· neutral

Git Knows Who. AI Knows What. Nobody Knows Why.

A developer has highlighted a growing crisis in software engineering: while AI can now generate entire features and Git can track who wrote code, the reasoning behind why code exists is increasingly lost. The problem is that critical context—requirements, discussions, design decisions, and AI conversations—lives scattered across separate tools like Linear, Slack, Notion, and Cursor, with no unified way to trace intent. The developer proposes a solution: a tool that lets engineers press "Ctrl+Y" on any line of code to see the full chain from requirement to implementation, preserving the reasoning that today's workflows consistently fail to capture.

read3 min publishedMay 31, 2026

Modern software development has achieved incredible things.

AI can generate entire features.

Editors can autocomplete your thoughts before you've finished having them.

Agents can open PRs while you're still reading the ticket.

And yet, despite all this progress, there is still one question capable of ruining a senior engineer's afternoon:

Why the hell does this code exist?

Consider this innocent little gem:

if (distance < 50) {
  return;
}

Looks simple.

AI can explain it.

Git can tell you who wrote it.

The PR can tell you when it was merged.

But nobody can tell you why it was added in the first place.

Was it reducing GPS noise?

Was it preventing duplicate events?

Was it added because thousands of devices were rapidly entering and exiting the same geofence?

Was it a workaround for a production incident that woke up three engineers on a Sunday morning?

We may never know.

Every mature codebase eventually turns developers into archaeologists.

You discover a mysterious piece of code and begin the sacred ritual:

Congratulations.

You have reached the end of the knowledge graph.

The only remaining documentation is a comment that says:

// Don't remove this

Thank you, mysterious engineer from 2023.

Very helpful.

The old workflow looked like this:

Human thinks
    ↓
Human writes code
    ↓
Human forgets why

The new workflow looks like this:

Human writes ticket
    ↓
AI writes code
    ↓
Human edits code
    ↓
Another AI refactors code
    ↓
Reviewer requests changes
    ↓
Code reaches production
    ↓
Everyone forgets why

We have successfully automated everything except remembering our decisions.

The funny thing is that companies already have mountains of documentation.

The requirement is in Linear.

The discussion is in Slack.

The design is in Notion.

The implementation is in GitHub.

The AI conversation is in Cursor.

The meeting notes are somewhere in a folder named:

Final_v2_Updated_Final_Real_Final

The problem isn't missing information.

The problem is that all the information lives in different universes.

Git is fantastic.

Ask Git:

Who changed this?

Git:

Bob.

Ask Git:

When?

Git:

February 14th, 2026.

Ask Git:

Why?

Git:

Commit message: "fix stuff"

Fantastic.

Outstanding.

Truly the pinnacle of human knowledge preservation.

We already have:

What we don't have is:

Ctrl + Y

Which means:

Why is this here?

Imagine clicking a line of code and seeing:

Reason:
Ignore GPS drift within 50 meters.

Requirement:
TRACKING-123

Discussion:
Customers reported phantom arrivals and departures.

Implementation:
PR #482

Generated by:
Cursor

Reviewer:
Sarah

Business Assumption:
Location updates within 50 meters are considered noise.

Now that's useful.

Because most bugs aren't caused by developers not understanding code.

They're caused by developers not understanding decisions.

For decades we've been obsessed with source code.

Then we became obsessed with documentation.

Now we're obsessed with AI code generation.

Meanwhile the most valuable thing keeps disappearing:

The reasoning.

The chain actually looks like this:

Requirement
    ↓
Discussion
    ↓
Decision
    ↓
AI Generation
    ↓
Code Review
    ↓
  Code

Today we only preserve the last two steps.

Then six months later we hold a meeting to rediscover the first four.

What if a VS Code extension could answer:

Why does this code exist?

Not by hallucinating.

Not by guessing.

But by building a traceable chain:

Requirement
    ↓
Discussion
    ↓
AI Session
    ↓
Code Review
    ↓
  Code

Click a line.

Press Ctrl+Y.

Get the story.

Not just the syntax.

People think AI-generated code is the next challenge.

I disagree.

The next challenge is AI-generated code with missing context.

Code can be read.

Logic can be reverse-engineered.

Intent is much harder.

And every year we're generating more code while preserving less reasoning.

That's a dangerous trade.

Git knows who.

AI knows what.

Nobody knows why.

And somewhere inside your production codebase is a line that nobody dares delete because the original author left three companies ago and the only documentation says:

// Trust me

Which, historically, has never caused any problems.

── more in #artificial-intelligence 4 stories · sorted by recency
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/git-knows-who-ai-kno…] indexed:0 read:3min 2026-05-31 ·