Four years ago we started building an HR platform

HR Pulsar — Mon, 18 May 2026 10:56:41 +0000

Not because HR-tech looked exciting. Mostly because every company we talked to had the same setup:

grades in Excel
360 reviews in Google Forms
competency matrices buried in Confluence
“performance systems” that nobody wanted to open twice

And eventually: “Let’s just build our own thing.”
So we did.

Then this year we made a bigger decision: we rewrote the platform from the ground up — and open sourced it.

Meet HRPulsar

An open source talent management platform for teams where every employee already works with AI tools.

Most HR systems still model companies like it’s 2015: employee → position → annual review.

But that model is breaking. Roles change every quarter. Skills decay faster than job titles. Half the team uses AI agents nobody tracks. And “career frameworks” are often PDFs with better branding.

We built HRPulsar around a different assumption:
Skills are the stable unit. Not positions.

So the platform is competency-first at the core: competency graphs, grade systems tied to behavioural anchors, 360 assessments, internal talent marketplace, development plans, AI fluency tracking, AI workforce registry
Yes, AI workforce registry.

Because “Shadow AI” is becoming a real operational problem: teams use 15 different AI tools, security doesn’t know about half of them, and HR systems pretend none of this exists.

So we added:

AI tool registry
employee ↔ AI tool assignments
oversight levels
workflow ownership
audit trails
workforce maps for hybrid teams (human + AI)

Useful for compliance & security. But mostly because companies should probably know which AI systems are already part of their workforce.

Technically, HRPulsar is

FastAPI
Next.js
PostgreSQL
Redis
200+ API endpoints
~1000 automated tests
Docker-first deployment
AGPLv3

No “book a demo”. No sales call before seeing the product.
And one thing we decided early: HRPulsar will stay free for individuals and small teams.

Every workspace gets a monthly pool of renewable credits. Enough to run the core platform without turning “try the product” into a budgeting discussion.

And yes — self-hosting is fully supported. Because HR data is sensitive.
And “trust us with your entire workforce structure” is a pretty big ask from a black-box SaaS vendor.

We’re preparing the public GitHub release right now.
*The repository goes live at the end of May. If you want early access, release updates, or just want to watch the project evolve — join the waitlist.

LLMs are useful ONLY for competency matching and draft recommendations

They are not qualified to decide someone’s promotion. We’re very explicit about that in the product.

LLMs are useful for competency matching and draft recommendations.
They are not qualified to decide someone’s promotion. We’re very explicit about that in the product.

Current limitations?
We don’t cover recruiting yet. ATS/recruiting is on the roadmap, but not today.

Would genuinely love feedback from developers building internal tools, people running self-hosted infrastructure, HR engineers, teams experimenting with AI-heavy workflows... and especially from anyone who has ever tried to manage competencies in Excel. We know you exist.

try it on website

Why AI adoption fails inside companies

HR Pulsar — Thu, 07 May 2026 12:15:48 +0000

A company buys ChatGPT Enterprise. Then:

marketing uses it for copy
one engineer automates half their workflow
another refuses to touch it
managers say “we should use AI more”
nobody knows what “more” means

Three months later, leadership asks the inevitable question: “So… are we actually getting ROI from this?”. Silence. Not because AI failed.
Because adoption inside companies is mostly random. And random systems don't scale.

Most companies approach AI rollout like this:

Buy tools
Announce initiative
Hope employees figure it out

That works for maybe two weeks. After that, you get what every company gets:

inconsistent usage
inconsistent output
no shared standards
no visibility into who’s actually effective with AI

Useful? Sometimes.
Measurable? Not really.

The uncomfortable part: most companies still evaluate people like AI doesn’t exist.
Performance reviews ask communication, ownership & collaboration.
Fine. But now we also need to ask:

can this person delegate effectively to AI?
can they verify AI output?
do they know when not to use it?
are they faster because of AI — or just noisier?

Because there’s a difference between “uses AI” and “works effectively with AI” is enormous.
And this is where most AI adoption projects quietly break. Not on infrastructure. Not on tooling. On management.
No competency model. No measurement system. No shared definition of “good AI usage” - just licenses and optimism.

The weird part is that companies already solved this problem once.
We don’t say: “Everyone has Excel now, good luck”. We train people, define expectations, measure proficiency.
AI will end up the same way. Except the impact is bigger.

At HRPulsar, we’ve been thinking about this a lot. Not as “AI replacing employees”. But as: how do you systematically measure and develop AI fluency inside teams?
And honestly, we don’t think the industry has good answers yet.

Especially for:

role-specific AI competencies
measuring real usage quality
separating employee contribution from AI contribution One thing already seems obvious: buying AI tools is easy, building an organization that actually knows how to use them is the hard part.

Curious how other teams are handling this right now. Do you actually measure AI adoption in any meaningful way — or is it still mostly vibes?

DEV Community: HR Pulsar

Four years ago we started building an HR platform

Meet HRPulsar

Technically, HRPulsar is

LLMs are useful ONLY for competency matching and draft recommendations

Why AI adoption fails inside companies