VALUE SPEAKS
FOR ITSELF

How to Replace Enterprise Software
and Become Your Company's Most Valuable Builder

CALEB FEDYSHEN
Software Engineer in AI  ·  Plains Midstream
hyprpixl.ca

Follow along on your device.

URL

hyprpixl.ca/talks/c2c-2026

All 35 slides.
Navigate at your own pace.
Arrow keys or swipe.

A retail bank was about to start charging us this
to access our own historical cheque data.

Billed per request.  We make hundreds of requests a day.

I built the replacement.

Four messy data sources  ·  Two banks  ·  One search bar

Every company in this room…

• …is paying for software that does 80% of what they need and 20% of what they don't.
• The procurement cycle to replace it: six months.
• The contract lock in: three years.
• The UI was designed by a committee in 2009.

And somewhere in that company, someone is keeping a spreadsheet
to work around the tool you're paying $50k a year for.

How to spot unmet software needs

• Excel doing database work. The universal tell.
• We have to manually… Every dev's ears should perk up.
• Software everyone hates but nobody challenges. Pain that's been accepted because replacing it seems impossible.

Build vs. Buy: the honest version

If a $20/month tool already exists and solves your problem,
buy it. Don't write code.

But build vs. buy now often means:
hire another company to do $300,000 of custom development work
because the people making the call
don't realize someone in house could build it for a tenth of that.

A competent developer can build something better, cheaper, and more tailored to the exact problem in weeks.

Most companies don't even know that's an option.

That's the opportunity.

Pain → Build → Outcome  ·  Numbers first.

ChequeViewer

• Bank sunsetting old image viewer. New one: billed per request.
• We pull hundreds of records/day for fraud, legal, and audit.
• Four data sources: zips of images, XML metadata, PDF pipeline, Excel files.
• Two banks. Four formats. Terabytes of zips.

Multithreaded extraction → queryable database → Flask app → Databricks App

Six million cheques.
One search bar.

saved vs. bank's new pricing

Deal Capture

My company makes money by moving oil through pipelines.
That starts with deals. Deals start as emails.

~10 hours/week saved  ·  error reduction  ·  one source of truth

Before: inbox + spreadsheet + copy paste + missed fields.
After: one place. Same data. Everyone. All at once.

Here's a problem every company has
and almost nobody has solved.

Contracts, agreements, SOWs, sitting in SharePoint folders.
Nobody knows what's inside any of them.

Legal asks: Find every contract with a renewal clause expiring in Q2.

Somebody starts opening files one at a time.

Google search, but for your contracts.

The same pattern happens all the time.

Build cost: a few weeks of one developer.
Off the shelf equivalent: pick a number and add a zero.

Quick show of hands.

How many of you are in an internship right now,
still in school, or just a few years into your career?

What I'm about to share is some stuff that's worked for me,
and worked surprisingly well.
Take it for what it is.

Value Speaks for Itself

There's a common piece of advice in tech:

Ask forgiveness, not permission.

I don't love it. It assumes you're doing something that requires forgiveness.

The better frame: build something so good
that forgiveness was never on the table.

Most developers wait.

Enterprise procurement exists to manage risk on uncertain bets.
Your job is to make the bet so obviously good
it collapses that process.

When you show a director a working tool
that saves her team twenty hours a week,
nobody asks how you spent last Thursday.

Legitimacy comes from outcomes, not process.

Everybody has more time than they think.

The hours in a week where you could be building something high leverage
is much larger than the hours you're currently using that way.

Replacing a tool that costs your company a million dollars a year
is so disproportionate it almost shouldn't be possible.

But it is.

I started as an intern. Nobody handed me a spec or told me to build anything.

I just saw a problem, had access to some tools, and started building.

Flask, Power Apps, a Databricks notebook, whatever.
The tool doesn't matter. The output does.

How to find your first win  ·  How to compound it

The one habit that
compounded more than anything.

Finance. Treasury. Ops. Anyone who doesn't write code for a living.
After a few weeks: the seeds of a few good ideas
and some people I'd actually talked to.

Ask three questions. Every time.

• 1   What's the hardest part of your job?
• 2   What tools do you use, and what's frustrating about them?
• 3   What's the most stupidly expensive piece of tech you have to use?

The snowball question:

Who are 2–3 people
you think I'd learn something from?

• Warm intro. No cold outreach. Credibility travels.
• A clearer picture of what's actually broken.
• People will tell you their real problems if you just ask.

Complaints are a
product spec
in disguise.

Increasing your luck surface area

Luck surface area = the number of people who know what you can do.

• Ship fast. One deployed thing beats ten planned ones.
• Post about what you build internally and publicly. Your SVP is reading.
• Name wins with numbers. Saved $40k/month travels. Built a dashboard dies.
• Make the person you helped your sponsor. They have political capital. They'll spend it on you.

Hey, I saw your post about that app you built.
There's this thing in my workflow that's annoying.
Could I pick your brain about it?

That happens every single week once you're consistent.
You stop hunting for ideas. The ideas come to you.

Here's what to do this week.

1. 01Find one person from a different part of your company. Ask them the three questions.
2. 02Listen for the manual work. That's your opportunity.
3. 03Sketch the replacement. Don't pitch it yet.
4. 04Build a slice. Show one person.

Every million dollar tool replacement starts with
one coffee chat and one prototype.

The companies that win aren't the ones with the biggest software budgets. they're the ones with developers who refuse to accept that the current tool is the only option.

You don't need permission. You need a problem worth solving.

Thanks.

I'm Caleb Fedyshen.

Find me on LinkedIn or at hyprpixl.ca