Financial services, technology, manufacturing, government
AI coding agent on your own GPU
Which AI coding assistant runs on our own GPUs so source code never leaves?
Katonic gives developers an AI coding assistant that runs on the company's own GPUs, knows the company's coding standards and never sends code outside. Developers keep the tools they already use, and security gets one audited endpoint. HPE has published this use case as a solution brief for HPE Private Cloud AI.
At a glance
| Industry | Financial services, technology, manufacturing, government |
|---|---|
| Who it is for | CTO, Head of Engineering, platform engineering leads, CISO |
| Data that stays in place | Source code, prompts, completions and usage logs |
| Runs air-gapped | Yes |
| HPE material | HPE solution brief "Coding Agent on Your Own GPU" |
| Infrastructure | HPE Private Cloud AI, or HPE GPU servers in an HPE AI Factory design |
The problem
Developers in banks, manufacturers and government want AI coding assistants, but source code and coding standards cannot be sent to a hosted model.
Unapproved use of public tools is a security and intellectual-property risk. Blocking the tools loses the productivity.
How it works
- An open-weight coding model is served on the company's own GPUs.
- Developers' editors and tools connect to one internal endpoint behind the Katonic AI Gateway.
- The assistant is grounded on the company's coding standards and internal documentation.
- Guardrails and access policies apply to every request.
- Every request is logged, with usage reported per team.
On HPE infrastructure
On HPE Private Cloud AI
HPE Private Cloud AI is HPE's turnkey private AI system, co-engineered with NVIDIA. Katonic AI is a partner in the HPE Unleash AI program. On Private Cloud AI, Katonic connects to the models the system serves and adds the agent platform on top. The coding model is served by the system, and developers reach it through the Katonic gateway. This is the configuration described in the HPE solution brief.
On HPE GPU servers
Customers who build their own estate on HPE GPU servers, as part of an HPE AI Factory design, get compute, storage and networking from HPE. Katonic supplies the software layer above the hardware: open-weight models served on the customer's GPUs, the AI gateway, the agent platform and governance. The coding model, the gateway and the logs run on the company's GPU servers, including in environments with no outside connection.
More on both routes: Katonic AI on HPE.
Where it stands
Published by HPE as a solution brief. A recorded demo is available.
Questions and answers
Does any source code leave our network?
No. The model runs on the company's own GPUs and requests do not go to a public AI service.
Do developers have to change tools?
No. Developers keep the editors and tools they already use and point them at the internal endpoint.
Which models can it use?
Open-weight coding models that the company chooses and serves on its own GPUs.
What does the security team see?
One endpoint with access policies, guardrails and a log of every request, with usage by team.