Local AI for useful workflows and controlled data.

An interactive client-facing example: lead analysis, JSON extraction, summary, customer reply and local-documentation Q&A, with operational data kept in the beneficiary's controlled environment.

Lead analysisJSON extractionSummary and actionsCustomer reply

Controlled environment

Provider: loading
Model: -
Endpoint: local mock
Data: check env

Task

Model input

Result

lead-analyzermocklocal-ai-demo-mock

Generated demo output appears here.

What the customer sees

The demo shows useful company workflows, not only generic chat.

Lead analysis

Score, intent, budget, urgency and follow-up questions.

JSON extraction

Turns free text into structured data for CRM or applications, without unnecessary exposure.

Summary and actions

Reduces documents or emails to clear steps and risks.

Customer reply

Drafts a short professional reply focused on the next step.

Local Q&A

Answers from local documentation, useful for a private knowledge base and controlled data.

Demo architecture

The same interface can run mock, Ollama or a local/NPU endpoint, with processing kept under control.

Client browser

eINFO demo API

Task prompt

Mock/local/NPU provider

Controlled data

Text or JSON result

NPU connection

When the local runtime is ready, only the provider and endpoint URL change.

Demo config

EINFO_DEMO_AI_PROVIDER=local-http
EINFO_DEMO_LOCAL_AI_URL=http://127.0.0.1:8787/generate
EINFO_DEMO_LOCAL_AI_MODEL_ID=qwen2.5-1.5b-instruct-npu-int8

Runtime contract

{
  "prompt": "...",
  "task_id": "lead-analyzer",
  "model": "qwen2.5-1.5b-instruct-npu-int8",
  "max_new_tokens": 384,
  "temperature": 0.1,
  "response_format": "json"
}

Signed eINFO update

The demo also shows personalized updates verified by manifest, checksum and rollback.

Mode: demo

Protected local data

.env.localdocumente clientbaze de date