🛠️ Build vs. buy · An honest essay

"I'll just have Claude
build it this afternoon."

Honestly? It probably can. Every team has this conversation now, most of our customers had it too, before they found us. This is the long version of the chat: what the afternoon actually buys you, and what the next twelve months cost.

Read the three questions →← Back to BuddiesHR

~6 min read · Based on conversations with 80+ teams that built before they bought

🤖 “But Claude writes the code now”

AI made the build nearly free.
The build was never the bill.

We agree: Claude will write you a working kudos bot before your coffee is cold. The question was never “can we produce the code?”, it was “who owns it in month eight?”

What Claude compresses

✓Writing the v1, genuinely, an afternoon
✓Block Kit & Adaptive Card boilerplate
✓The Postgres schema and migrations
✓Even the unit tests

i.e. the part that was already the cheapest line on the estimate.

What still lands on you

→Being on-call when birthdays don’t post
→The Slack deprecation that lands in month 6
→Security review of code nobody on the team wrote
→HRIS credentials, syncs and conflicts, forever
→HR’s feature requests, every quarter
→Ownership, when the person who prompted it leaves

The cheaper the v1 gets, the easier it is to end up owning software no one wrote.

🛠️ Build vs. buy

Thinking of building this in-house? Three questions first.

Most of our customers considered a "quick Slack bot" before they found us. Here's the conversation we have with them.

Is there already a tool focused on solving this?

Building it once is the easy part. Maintaining it for three years is where these projects quietly die.

Timezones · Slack rate limits · Teams cards · GDPR deletion · HRIS imports

Do you have someone with the appetite to maintain it?

Spinning up a script in a sprint isn't the same job as owning a product for years.

Slack API deprecations · Docs for the next hire · The 3am birthday alert in São Paulo

What happens if that person leaves?

This is the question that quietly burns HR teams.

Code in someone's old .env · No retention policy · Employee data with no owner

Build, if

It's truly unique to your company
You have at least 1 employee committed to ownership
Maintenance is on a job description
It's strategic, a product, not a tool

Buy, if

It's a solved problem, kudos, surveys, birthdays, PTO
HR is the buyer, not engineering
You want it live this week, not in Q3
You'd rather your team focus on what's unique to you

Most teams who build it end up patching half-finished bots every month. BuddiesHR is what they were trying to build, without the maintenance bill.

Skip the build →

💸 The estimate vs. the bill

The estimate is always the v1.
The bill is everything around it.

We asked four engineering leads to compare what they pitched to leadership vs. what they actually spent. Composite, rounded honest.

The pitch

What it actually was

What no one budgeted

Estimate ×

"A weekend Slack bot"

5–8 days for a v1 that handles only happy paths

Time-zones, retries, Slack rate limits, error states, "what if the channel was archived?"

~3×

""We'll just store it in Postgres""

A schema, migrations, a back-office UI, GDPR exports & deletion

Permissions model, audit log, "who can see what" arguments in three meetings

~4×

""We'll add Teams later""

A near-full rewrite, Teams adaptive cards are not Slack Block Kit

Tenant install flow, MS Graph permissions, separate review by IT

~2×

"Maintenance is 1 hour a week"

Slack ships breaking changes ~quarterly; API deprecations land cold

On-call for HR tooling, internal Jira, the engineer who left

ongoing

📅 A year in the life

What actually happens
after the MVP ships.

Month 1
Excitement
Claude writes the MVP in a day. Slack messages get sent. Team posts a screenshot in #demos.
Month 2
First edge cases
Birthday at midnight UTC pinging the wrong country. Three Jira tickets. Still excited.
Month 3
Feature creep
HR asks: "could it also do anniversaries? And kudos? And a survey? And export to Notion?"
Month 4
Teams ask
A new acquisition uses Microsoft Teams. The bot doesn't. Estimate doubles.
Month 6
The Slack API change
Block Kit deprecation. Half a sprint to migrate. No new features that month.
Month 8
The original engineer leaves
Knowledge transfer is "the README and one Loom." Velocity on the bot drops to zero.
Month 10
Security review
IT asks where employee data lives. Answer is "in a Postgres on Heroku." Meeting follows.
Month 12
The conversation
"Should we just buy something for this?" Most teams arrive here within the first year.

🧊 The iceberg

What "just an engagement app"
is actually made of.

↑ What gets pitched

•Send a Slack message on a birthday
•Post a kudos card in #shoutouts
•A weekly pulse question

≈≈≈ waterline · everything below is also you ≈≈≈

Identity

·SSO with Slack & Teams
·Multi-workspace + multi-tenant
·User lifecycle on HRIS sync
·De-provisioning on offboarding

HRIS sync

·BambooHR / Personio / Rippling / HiBob / Workday connectors
·Field mapping UI
·Rate-limited backfills
·Conflict resolution on overlapping sources

Scheduling

·Cron in N time-zones
·Skip company holidays per country
·Quiet hours
·Retry on Slack 5xx & re-deliver

Content & UI

·Block Kit templates
·Teams Adaptive Cards (different language!)
·Internationalisation in 6+ languages
·A back-office UI for HR to edit copy

Data & privacy

·Postgres + backups + PITR
·GDPR data export & right-to-be-forgotten
·EU vs US data residency
·Audit log for HR-visible actions

Reliability

·Status page
·Alerting on missed birthdays
·On-call rotation
·Slack & Teams deprecation tracking

Distribution

·Slack App Directory review
·Microsoft AppSource submission
·Marketplace screenshots, privacy policy, support email

Support

·"It didn't post" debugging
·Onboarding new admins
·Training new HR hires every quarter

BuddiesHR is the iceberg, productised, and split into nine apps so you only pay for the ones you turn on.

We built a kudos bot in a hackathon. It ran for 14 months. Then Slack changed Block Kit, the engineer who wrote it moved to platform, and HR was suddenly maintaining a Python script via Notion. BuddiesHR was cheaper than three days of that engineer's time.

Sam Okonkwo

Director of People Ops · 240-person SaaS

The fastest way to improve
your team's experience
is to install one app and see.

30-day free trial. Free forever under 10 employees. Public pricing after that. Uninstall in one click if it's not for you. No sales call in sight.