← → 翻页 · ESC 索引
Engineering Talk · 2026.04
Vol. 01
Notes from 6 weeks with Vibe Coding

Code is Cheap.

Now what?

The coding agent has quietly revolutionized how I work. The real question is no longer can we build it — but what are we choosing to build, and why.

Burt·Sainstore Engineering·For the team
An honest reflection · Not a tutorial
— 2026 —
Context · The Setup
02 / 22
6 weeks. Daily driver. No exceptions.

I've used Coding Agents
every day for 6 weeks.
Here is what I learned.

In this deck·5 Shifts·1 workflow·3 new expectations
Page 02 · The setup
— · —
Context · By the numbers
03 / 22
Same company, same engineer — two stacks.

Five years. Then 6 weeks.

Before · InventoryVue + InstaBIS
229K
lines
Vue 171K · JS 49K
3,753 commits · entire history
One team, four years
~3 per day average
After · sainops
266K
lines
TS 121K · Vue 142K · +3K
329 commits · 6 weeks
One PM, 6 weeks
~7.8 per day

LOC is a terrible productivity metric — and even this terrible metric can't hide what happened. Both figures exclude CSS, markdown, SQL, lockfiles, and binary assets.

Page 03 · The receipts
InventoryVue · sainops · April 2026
Context · 6 weeks of sainops
04 / 22
Commits · a day-by-day picture

329 commits. 6 weeks.

Fun facts
Busiest day · Mar 27 (32)
Peak hour · 2pm (35)
Night owl · 33 commits, 12→1am
Commits per day · Mar 15 → Apr 28
max 32 · Mar 27
Mar 15 Mar 25 Apr 08 Apr 28
Total commits
329
across 41 active days
Peak day
32
Mar 27 · one heck of a shipping spree
Active days
41/42
shipped code 98% of the window
Page 04 · 6 weeks of commits
329 commits · sainops
Act I · Five Core Shifts
Act I · 05 / 22
Act I

Five Core Shifts.

What actually matters when code stops being the bottleneck.

Act I opening
— · —
Act I · Shift 01
06 / 22
01
Shift 01

Drop your chat box.

The IDE is the past. Embrace the CLI.

If you're still copy-pasting answers from a chat window, you haven't started yet.
— The shift that matters most
Page 06 · Shift 01
Think like a user
Act I · Shift 02
07 / 22
02
Shift 02

Build fast feedback loops.

Fast customer
feedback.

Ship → measure → iterate. The agent writes code. Your users tell you if it matters. Close the loop in hours.

Agent self-
improvement.

Invest in dev infrastructure that allows agent feedback loop.

Page 07 · Shift 02
Fast feedback
Act I · Shift 03
08 / 22
03
Shift 03

Dream big.

Ideas that lived in the "too expensive, skip it" column a year ago now belong on the roadmap. The cost curve flipped — our ambition has to flip with it.

The right question isn't "what small task can I delegate?" but "what would I attempt if implementation cost was no longer the constraint?"
Page 08 · Shift 03
Dream big
Act I · Shift 04
09 / 22
04
Shift 04

Plan accordingly.

Without a plan, ambition just means wrong code, faster.

Dream big, but align first — or you'll ship 2,000 lines of confidently wrong code.

Page 09 · Shift 04
Foundation
Act I · Shift 05
10 / 22
05
Shift 05

More Thinking,
Not Less.

When the agent writes the code, human judgment becomes the constraint. We shift from can we build this to should we — identifying gaps in the spec, catching logical slips before they ship, and setting the right direction.

This is the job now. Questioning, scrutinizing, deciding — not producing.

Page 10 · Shift 05
Human judgment
Interlude
11 / 22

One Takeaway.

Act II
Act II · 12 / 22

Ask not what your agent
can do for you—

ask what you can do for your agent.

Act II opening
— · —
Act II · The Shift
13 / 22
Ask what YOU can do for your agent

Three things that matter.

01 · Context
Context
matters.
02 · Harness & Tests
Rails the
agent rides on.
03 · Judgment
Your taste
is non-negotiable.
Page 12 · What you give the agent
Context, Harness, Judgment
Act III · New Expectations
Act III · 14 / 22
Act III

The bar just moved.

Three expectations I'm holding myself — and this team — to, starting now.

Act III opening
New expectations
15 / 22

The world is moving
FAST. We should too.

— · —
Act III · Expectation 01
16 / 22
01

Everybody is a product
BUILDER.

Be an individual contributor (again).

Page 16 · Expectation 01
Act III · Expectation 02
17 / 22
02

"It works" is the floor,
not the ceiling.

Ask yourself — is it the right solution, or just a solution?

Automated tests are the norm, not the nice-to-have.

Page 17 · Expectation 02
Act III · Expectation 03
18 / 22
03

Builders are
designers now.

Design thinking must be embedded in product development. Know what to build, why, and what good looks like.

Page 18 · Expectation 03
19 / 22

Code is free.
Taste isn't.

Anyone can generate code now. Fewer people can tell you which code deserves to exist.

Bad taste compounds.

Page 19 · Code is free
— · —
20 / 22
One more thing

Code is Cheap.
Show me something else.

Page 20 · The question
— · —
the beginning is the end is the beginning
21 / 22

It is the best
time to create.

Thanks.·Questions?·Burt · Sainstore
Page 21 · The beginning
— End —
22 / 22

Additional Readings

Harness Engineering
openai.com/index/harness-engineering/
Claude Code Best Practices
code.claude.com/docs/en/best-practices
Using Claude Code: Session Management & 1M Context
x.com/trq212/status/2044548257058328723
Must-have skill
obra / superpowers
github.com/obra/superpowers
Page 23 · Additional readings
— End —