Governed Analytics Inside Your Own Walls: Four Ways to Deploy Chat4ED
Most AI analytics tools send your question, your schema and often your data to a model hosted by someone else. For a bank, a hospital or a government team that is often not allowed, whatever the vendor promises. The rules are clear: customer data, patient records and financial ledgers stay where the regulator can see them.
Chat4ED was built for those teams. The same product runs in four places, and the difference between them is simply how much leaves your network.
1. Chat4ED Cloud
The fastest start. We run and update the platform for you, and you can be live in days. It suits teams that already use managed cloud services for their data and want answers before they invest in infrastructure.
2. Private cloud
Chat4ED is deployed into your own AWS, Azure or GCP account. Your data never leaves your VPC, and you hold the encryption keys and the logs. Your cloud security controls apply to Chat4ED exactly as they do to the rest of your estate.
3. On-premise with local LLMs
Chat4ED runs on your own servers with locally hosted language models such as Llama, Qwen or Mistral, or a model you have fine-tuned yourself. No question, schema or result is sent to an outside AI provider. You pay for hardware instead of per-token API fees, which also makes heavy everyday use predictable to budget.
4. Air-gapped
For the most regulated environments, Chat4ED runs with no internet access at all. Installation and updates are delivered offline.
What stays the same everywhere
Wherever it runs, every answer goes through the same controls:
Multi-model verification: one model writes the SQL, another reviews it and a third validates the result. If they disagree, Chat4ED asks you instead of guessing.
Answers that show their work: every answer comes with the business definitions it used, the SQL it ran and the tables behind it.
Access rules respected: people only see the rows and columns their role allows, sensitive fields are masked, and every query is logged for audit.
Choosing the right option
Start with what your data is allowed to do. If it may sit in a managed cloud, Chat4ED Cloud is the quickest route. If it must stay in your own cloud account, choose private cloud. If no outside AI provider may see it, run on-premise with local models. And if the network itself must stay closed, go air-gapped. You can also start in one model and move to another later, because the product is the same everywhere.
