Your data stays with you
Kepler runs on your machine. Bring your own model keys, free on every plan.
Kepler opens to everyoneon 20 November.
07days
15hours
23min
36sec
Developers got their AI IDE the day their repos held still. The people keeping production alive never did. Kepler is the IDE for the ones on call.
Join the operators already insideWhy Kepler
A coding agent works because the repo holds still. An operator acts on a living world that forgets the moment you look away.
Kepler builds the two things ops never got for free.
A world model
That deepens with every incident.
A standing presence
That watches, and wakes itself when something moves.
Every finding shows its proof
The commands it ran, what it saw, and what it is only guessing.
It suggests. You decide.
It asks before it changes anything, and can never loosen its own limits.
It remembers, for you
What you teach it stays with you and makes the next incident shorter.
136 days,first commit to launch.
Built in the open with the operators who would use it, and changed every time they told us it was wrong.
- First commit
- First alphav0.0.3 for a few operators.
- Beta cohort76 operators on real systems.
- Hack Day Pune146 people in one room.
- Opens to everyone
600+downloads
220+daily active users
118Discord cohort members
300+ operators shaped it.This launch is theirs too.
The beta ran in the open: the cohort, office hours, meetups and a full day of building in Pune. People ran Kepler on real systems, broke it, and told us what scared them.
7,567requests in 12 days
3,095on the busiest day
86%asked for the fast model
168public posts about what they built
You said. We changed.
It says it's reconnecting when nothing is wrong.
Fixed the next day. We rebuilt how the app talks to the agent, so a busy session no longer looks like a lost one.
Too many approval walls. I can't work like this.
Approve once, keep the rule. Kepler remembers what you allowed, and still asks for anything new.
My model keys are sitting in a readable file.
Fixed the next day. You were right, and we said so.
When something breaks I want an answer now, not an essay.
Fast by default. 86% of requests went to the fast model, so it became the default and depth is one step away.
Every light is someonewho made it better.
Most AI tools for operations want your production data in someone else’s cloud. Kepler runs on my machine, works with the model I pick, and drives a real browser through my dashboards. It does the legwork and I never leave my flow.
Mridul GainSMTS, Platform9A deployment was crashlooping in production on out-of-memory. I gave Kepler the heap dump and the code, and it pointed to the exact place causing it.
Lokesh WaranEngineering, Cimba AII showed Kepler to our senior VP of platform infrastructure. He was impressed and told me to keep going. Now I’m bringing the wider team on.
Jason PowellSenior Infrastructure Engineer, SkillableIt feels like talking to a chat agent, so there was nothing to learn. It just works.
Chindu ReddyAI engineerPeople who went out of their way
And everyone on the Discord
Join the Discord- SHShubham
- IMImran
- BLBloom
- RØRøháñ
- PRPrasenjit
- MAMaverick
- RARashi
- MEMEHUL
- SASaurabh
- SFSh3r!ff
- NDNandini Deshmukh
- GCground_control🛂
- SOsoham
- AMAdil Majeed
- VSVineet Srivastav
- GAgaurav
- SGSachin Govind
- HIhim7797
- KAKanva
- GCGaurav - CPO : MindHub
- KSKundan Singh
- PUPuran
- JKJayakumar k k
- SWSwastik
- BDBikram Debnath
- ELElmetaDor
- SHShiv
- RORoy
- SKSuraj Kumar
- SHSaurabh Hirani
- VKvinay kornapalli
- RARajneesh
- VSVivek SRE
- LOLokesh
- DIDipti
- SAsarkar.arthur
- DODosaDynamite
- ISIshant
- RBRandy Bias 🇺🇸🇵🇭🇸🇰
- JAJaideep
- SBSarthak Basu
- AMAmana
- SSSonu Singh
- MAmahala
- RAramiyengar
- RMRohan Mainali
- KAKage
- BHbhakt
- TVTitan valker
- PApacifer26
- ASAadish Sanghai
- ·🧋
- KAKartikey
- RARaghav
- SSSoul Surfer🌄
- SWSamir Wankhede
- S0shruti_01
- AKAjay Kumar
- IRirbansin
- CPcheerful_panda_54306
- GAGayatri
- ADAnirudh.D
- KAKaif
- HEHemant
- ARAryan
- TATalin
- NUnuke
- SIsimpliplay
- PCpropaine can
- DLDark Legend
- SRsridhark
- SOsohamdev77
- WIwityliti
- JOJohnP
- SOSourav
- PIpithagora
- VIvikas
- KUKubeNinja
- KAKashyap
- GOGojo2811
- ASash__
- CScsdeepak2005
- ACAaradhy Chinche
- KAkaysin
- EDedward007
- NENetworksGT
- CRChindu R
- PRPrakash
- SNskg-net
- PTPramod Thombare
- JPjpisgeek
- VSVivek Shukla
- SISimha
- SJSuz_Jailbreak
- AJAtharva Joshi
- PNParth Nerkar
- T2Tech@2005
- HEhemanthbalaji.
- VAVaishnav
- RMRishi Mondal
- WIwipphanp
- YOYoga
- MAManthan
- KVKaran Verma
Built in the open.With the people on call.
I have spent most of my working life keeping other people's systems alive. Coding agents got a perfect world model for free. The rest of us got dashboards at 3am. Kepler is my attempt at the other half: an agent that learns your infrastructure the way a good colleague does, and asks before it touches anything.
Hundreds of operators used the beta and shaped what it is today. On 20 November it opens to everyone. It ships every week and it is not finished. If something breaks or feels wrong, tell me on the Discord. I would rather hear it now than after it is polished.
Priyank UpadhyayFounder, RubixKube
Free to keep.$20 when you want all of it.
Free is a real tool, not a demo. Your own model keys work on every plan.
Free
Forever$0
For one machine
- Chat, terminal, browser and editor
- Memory, skills and slash commands
- MCP and the CLI
- Up to 3 watchers
- Your own model keys
Pro
$20a month
or $16 a month, billed yearly
- Everything in Free
- War rooms and topology
- Automations
- Remote access and channels
- Unlimited watchers
- Hosted models included
- 3 devices
AI credit
One-timeAdd credit
Buy once. No subscription.
- For when you reach your weekly limit
- Use it on any hosted model
- Lasts a year
Every new account starts with 14 days of Pro, hosted models included. No card. On call as a team? Talk to us.
It's time somethingstayed up with you.
Opens in 7 days, 15 hours and 23 minutes.