Klique is the control plane for enterprise AI. Govern every model, agent, GPU, and dollar across the stack you already own. Visit us at https://nitter.cf/t.co/dIi02euWVN
Joined April 2018
- Tweets1.2K
- Following528
- Followers3.7K
- Likes406
Pinned Tweet
ClearML Enterprise is now Klique, the control plane for enterprise AI. It adds an LLM gateway, model routing, and agent controls so IT can see what’s running, on which GPUs, and at what cost. ClearML open source stays free. CEO Moses Guttmann on why:
klique.ai/resources/content-…
#SovereignAI fails the moment a developer decides a public API is faster than your private GPU cluster.
Klique on @SUSE AI Factory with @NVIDIA lets devs launch approved models in 1 click & logs every call by user and model.
Find us at booth #7054 through Oct 22! #GTCBerlin.
How Klique keeps teams on the governed path when it’s running on @SUSE AI Factory with @NVIDIA:
klique.ai/resources/content-… #SovereignAI #GTCBerlin
Here's how Klique gives that agent its own identity and budget, and lists it where the rest of the finance team can run it:
klique.ai/resources/content-…
#AIGovernance #AIAgents
“Half my team was building agents on their laptops with Claude Code and personal API keys.”
That’s how one CIO summed up enterprise AI this year. When one of those developers leaves, the record of which repos went to which model may go with them.
#ShadowAI #VibeCoding #AIGovernance #EnterpriseAI
What a governed path looks like, and why it has to be faster than going around IT:
klique.ai/resources/content-…
#AIGovernance #VibeCoding #EnterpriseAI
Every AI tool IT bans gets replaced by the next one within a week, and the spend moves to personal API keys where #FinOps can’t see it.
Give builders a sanctioned path that’s faster than going around IT, with every request tied to a user or agent and checked against a budget.
What this looks like for FinOps teams managing AI spend and tokenomics:
klique.ai/resources/content-…
#FinOps #AgenticAI #AIFinOps #Tokenomics
Next week at #NVIDIAGTC Berlin, we debut as Klique (formerly ClearML): the control plane for enterprise AI. Runs on the GPUs you already own, on-prem or air-gapped. Booth #7054, Oct 20-22. Can’t make it? Request a demo: klique.ai/demo
So what does Klique’s control plane do? Picture an LLM gateway that sends each prompt to the cheapest model that can handle it, your GPUs scheduled as one pool, and spend you can check per project. It runs on the infrastructure you already have. More here:
klique.ai/resources/content-…
We're here at #DODIIS! Here are some scenes from yesterday's build-out.
Wondering what it takes to prove control and adherence to sovereignty? Come chat with us!
You can find us in Booth 742 & the @Carahsoft Partner Pavilion
Owning the infrastructure is the easy part of AI sovereignty. Proving you control it, on demand, is where things get real for most agencies.
If that's a question on your team’s mind, come find us at #DoDIIS Worldwide next week in Tampa, August 9-12.
Next week the team will be at #DoDIIS in Tampa. Find us in Booth 742 and the @Carahsoft Partner Pavilion.
#DoDIIS #AISovereignty #DefenseAI #NationalSecurity #ClearML #AI
Paul Garcia and Alex Paquette are at @AMD #AdvancingAI 2026, talking GPU infrastructure and more all day.
If you're running AI workloads across on-prem, cloud, or both, stop by and ask about the ClearML Infrastructure Control Plane. It's built to keep your GPU clusters fully utilized and your costs under control, no matter where the hardware lives.
Find us on the show floor!
#AMD #GPU #AIInfrastructure #CostOptimization
At @AMD #AdvancingAI today and tomorrow? Come by the ClearML to see how we handle the last mile of an AMD Inference Microservice (AIM) deployment.
AIMs ship pre-tuned for AMD GPUs & CPUs. ClearML handles access control, autoscaling, tenant isolation, and observability once it's live.
NERSC, the mission HPC center for the U.S. DOE Office of Science, is using ClearML across Perlmutter, the upcoming Doudna supercomputer, and the American Science Cloud. Slurm-native orchestration, HPO at scale, LLM serving, and a NERSC-built MCP server that lets agents drive ClearML directly.
clear.ml/blog/inside-nersc-a…
#AIInfrastructure #HPC #ClearML #AIForScience #NERSC @clearmlapp @NERSC
Thank you to everyone who visited the ClearML team at Booth A11 during ISC 2026 in Hamburg!
It was an amazing week of conversations around HPC, AI infrastructure, GPU orchestration, and scaling enterprise AI workloads in production environments.
We loved connecting with partners, customers, researchers, and innovators from across the global HPC and AI community.
Thank you to everyone who stopped by to join our conversations—we’re excited for what’s ahead. See you at the next event!
#HPC #ISC2026 #AIInfrastructure #EnterpriseAI #MLOps #GPUComputing #ClearML @clearmlapp
Most enterprises still size AI infrastructure for training, but that’s not the workload anymore. It is now ~2/3 of AI compute, and prefill/decode, KV cache, and reasoning tokens break old capacity math. The full breakdown:
clear.ml/blog/inference-is-t…
#AIInfrastructure #Inference #GPU #GenAI #CapacityPlanning #ClearML
