KMD
@kentuckymd64early
Taking longer than usual. Trying again shortly
Project
Project J.E.N.E.S.I.S
A place for developers to manage and see AI-supported software systems easily.
Taking longer than usual. Trying again shortly
Founders, what's the fastest way to get your first 10 paying users?
Taking longer than usual. Trying again shortly
microservices solved the deployment problem.
they created a comprehension problem nobody budgeted for.
your org chart and your architecture diagram are different documents now.
they shouldn't be.
Taking longer than usual. Trying again shortly
made a video showing what Jenesis actually looks like (shot and edited in one day so its shit), might be worth watching youtu.be/PBndt3G-aeY?si=aGUapr29v8XU0Gqd
Taking longer than usual. Trying again shortly
The incident lasted 23 minutes. About 18 of those were spent figuring out what changed. The actual fix took 5 minutes. If those numbers feel familiar, that's what I'm trying to fix.
Taking longer than usual. Trying again shortly
the hardest part right now isn't building Jenesis,
it's stopping myself from adding more features and trying to make it look futuristic and cool as fuck.
I want a sexy AI assistant and build the coolest possible thing
but first I need to answer a much simpler question:
does anyone actually want this?
I need to be patient.
Taking longer than usual. Trying again shortly
thing I got wrong in the first version:
tried to make Jenesis smart with automatic correlation, anomaly detection, etc.
scrapped it.
the value isn't the algorithm,
the value is having everything in one timeline so you can correlate
tools that try to think for you during incidents are usually wrong at the worst moment (my justification because I'm still figuring out how to make it automatic lol)
Taking longer than usual. Trying again shortly
phase 2 is done what that means:
- topology definition (YAML + UI) ✓
- event ingestion via webhooks + github ✓
- unified timeline ✓
- architecture-aware incident view ✓
now I just need to actually get engineers to try it on real systems.
Taking longer than usual. Trying again shortly
first look at the Jenesis dashboard. unified timeline on the left, service topology on the right. scrub back through events and watch the graph light up where things changed. ui looks like shit but the core idea is working!
Taking longer than usual. Trying again shortly
decided to go with manual topology definition instead of auto-discovery
mainly because:
auto-discovered graphs are only accurate during normal operation
during an incident, they're wrong exactly when you need them most
manual is dumber. manual is better for this use case ig.
Taking longer than usual. Trying again shortly
Jenesis build log — week 6
built: unified timeline view
broke: event ordering with millisecond-resolution timestamps
learned: clocks in distributed systems are a lie
next: adding sequence IDs to the event schema
still don't know if anyone will use this but the problem is real
Taking longer than usual. Trying again shortly
hey I'm KM, building Jenesis, its a self-hosted tool that correlates operational events (deploys, alerts, config changes) into a unified timeline so engineers can reconstruct incidents faster instead of manually checking 5 tabs.