I am the founder of SprintPulse. Like many of you, I used to dread retrospective meetings. They often turned into a repetitive cycle where we wrote down the same issues every sprint but never actually fixed them. "Better communication" was on our action item list for months.

I built SprintPulse to fix that loop. It is a tool designed not just to collect feedback, but to make sure it leads to real change.

When I looked at the tools available, I found they fell into two camps.

On one side, you have generic whiteboards like Miro or FigJam. They are great for free-form drawing, but they are too open-ended for structured retros. They don't help you track history or remind you of what you promised to fix last month.

On the other side, you have specialised retro tools. I found that while these solve the problem, they often feel outdated. The UX can be clunky, they are often loaded with complex settings you don't need, or they are just too expensive for smaller teams to justify.

I wanted to build something in the middle. A tool that has the structure and insights you need (for which AI isn't just a "me too" buzzword), but with a modern interface that doesn't require a tutorial.

- AI Analysis: Instead of just counting votes, the AI reads the feedback to spot trends and suggest action items. If the team is getting burnt out or if specific topics keep appearing without resolution, the system flags it. - Accountability: This was the main thing I wanted. SprintPulse tracks your action items from previous sprints. If you ignore a problem for three weeks, the tool brings it up. It forces you to either fix it or consciously drop it. - Clean UX: I focused on keeping it fast and lightweight, and enjoyable to use. You just log in and start the meeting.

I believe smaller teams shouldn't have to pay just to run a basic retro.

- Free Plan: We have a solid free plan for teams up to 10 users. It includes 3 retros and the core features. No credit card required to sign up. - Paid Plans: If you have a larger team or want deep historical analytics and insights, integrations with Linear, Jira and Slack, we have paid tiers.

For the HN community, I am offering 40% off any paid plan (recurring, forever) with the code HACKERNEWS.

I would love for you to take a look and tell me what you think.

Link: https://sprintpulse.io/

Thanks!

Comments URL: https://news.ycombinator.com/item?id=47249468

Points: 1

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=46395855

Points: 4

# Comments: 0

Comments URL: https://news.ycombinator.com/item?id=42848339

Points: 3

# Comments: 0

It uses k3s as Kubernetes flavor and Hetzner Cloud as provider. For this test I used extremely high concurrency so the tool hung twice in the middle of the process because I was hitting the Hetzner API too hard, so I had to interrupt it and continue.

Excluding the time it paused/hung due to the API, I calculated around 11 minutes total for the cluster creation. This includes:

- creating all the resources (cloud instances, firewall, load balancer for the Kubernetes API)

- deploying k3s to the control plane (3 masters) and the 297 worker nodes

- installing the Hetzner Cloud Controller Manager, to link K8s nodes to the cloud instances as seen by the Hetzner control panel and be able to provision load balancers out of the box

- installing the Hetzner CSI driver, to be able to provision block storage volumes out of the box

- installing the Cluster Autoscaler, to allow configuring autoscaling node pools

- installing the Rancher System Upgrade Controller, to handle k3s upgrades very easily

I believe this is a world record, or at least I have never heard of a tool managed or not to create clusters that comes even close to mine in terms of speed. Am I wrong? I would be curious to hear if there is a tool even faster than mine.

Of course not many people need to create clusters with 300 nodes from the get go, but it was a fun experiment. In the version I will release in the coming few weeks (v2.0) I will likely limit concurrency to more sane levels to reduce the risk of hitting the Hetzner API too hard, although most people will create small clusters to begin with. During my testing, I was able to create a cluster without hanging etc with max 100 nodes. Beyond that it almost always hangs and there are retries to create/power on cloud instances. So yeah I am going to limit the concurrency just in case to perhaps 10 or 20 servers per time for real life use.

What do you think of this? Let me know! I am curious to hear feedback or any comments on the subject :)

Comments URL: https://news.ycombinator.com/item?id=40149197

Points: 1

# Comments: 0

Everything is in a Github discussion at https://github.com/vitobotta/hetzner-k3s/discussions/296. A huge thank you in advance if you can help with this!

Comments URL: https://news.ycombinator.com/item?id=38449603

Points: 2

# Comments: 0

If you are already familiar with this tool, I'd love to know how it's worked for you so far. :)

Comments URL: https://news.ycombinator.com/item?id=37969307

Points: 4

# Comments: 0

The main change in this release allows creating node pools with any number of nodes - you are limited only by the caps on your Hetzner account. This update makes it possible to create pretty large clusters for cheap

https://github.com/vitobotta/hetzner-k3s

Looking forward to any feedback, comments or questions :)

Comments URL: https://news.ycombinator.com/item?id=35581701

Points: 3

# Comments: 1

Comments URL: https://news.ycombinator.com/item?id=34716306

Points: 1

# Comments: 0