FAQ
Questions to settle before you start.
Practical answers about privacy, cost, IT ownership, support, and scope.
Are you selling an AI app?
No. InHouse Compute is a deployment partner for private AI infrastructure. We help design, deploy, document, and support customer-controlled environments.
What happens after I send an inquiry?
We review it and reply within one business day. If there is a likely fit, we invite you to a free 20-minute Initial Fit Call to clarify your goals and next step. If you proceed, the assessment includes a working session and written recommendation.
Will this replace our cloud AI tools?
Not for every task. Frontier cloud models may still be stronger for some general reasoning work. In-house AI is strongest when privacy, document control, predictable infrastructure costs, offline access, or vendor independence matter.
Will this eliminate token bills?
For local workloads, you are not billed by an outside model provider per token. You still have infrastructure, maintenance, upgrade, power, and support costs.
Does our data leave our company?
The default goal is to keep approved documents and AI workloads inside the customer-controlled environment. Any remote access, hosted dependency, or external model use must be reviewed and approved by the customer.
What hardware do we need?
It depends on model size, document volume, concurrency, latency goals, and budget. We start with a readiness assessment and recommend a tier that can be validated with benchmarks.
Can our IT team maintain it?
Yes, that is the design goal. We provide admin handoff documentation, backup procedures, update notes, and optional maintenance support.
Can you connect it to company documents?
Yes, for approved document collections. Complex permissions, large document estates, and regulated data need additional scoping.
Do you choose the model for us?
We recommend a model path based on the use case, hardware, license terms, customer policy, and benchmarks where available. The final choice is documented and approved as part of the deployment record.
Do you support open-source models?
Yes, where model licenses, customer policy, and the use case allow. We are model-agnostic and document model choices for each deployment.
Can this run fully offline?
Some deployments can run offline after installation and model download. Offline operation affects updates, support, monitoring, and model refresh workflows.
Is this compliant with HIPAA, CMMC, SOC 2, or other standards?
We can design deployments that support customer-controlled data and infrastructure goals, but compliance depends on the full customer environment, policies, controls, and legal requirements. Formal compliance claims require a dedicated compliance review.
Do you offer support after deployment?
Yes. Maintenance and model refresh support can be scoped as an optional monthly plan with explicit access, response, and responsibility boundaries.
What happens when better models come out?
We document the current model, test candidates against approved use cases, and provide a controlled refresh path when better options are available.
Still have a question?
Send an inquiry first. If there is a likely fit, we will explain the paid assessment on a short initial call.