# Proposal: Private AI Deployment

**Prepared for:** [CLIENT NAME]  
**Contact:** [CONTACT NAME]  
**Date:** [DATE]  
**Proposal ID:** [PROPOSAL ID]

---

## Executive Summary

[CLIENT NAME] generates institutional knowledge faster than it can be organized. Decisions are delayed. Staff repeat searches through scattered documents, email threads, and the memories of colleagues who may not be here next year. The cost is measured in hours, errors, and missed opportunities.

We propose a sovereign AI deployment: a private, on-premise system that ingests your documents and institutional knowledge into an isolated knowledge base. Your staff query it. The system retrieves cited, reproducible answers from your own materials. No OpenAI. No third-party model dependency. No data leaves your building.

This proposal outlines the scope, timeline, and investment required to deploy [PROPOSED SCOPE] for [CLIENT NAME].

---

## Problem Statement

Every organization we work with describes a version of the same leak.

**[CLIENT NAME]'s specific situation:**

[SPECIFIC USE CASE]

[DEPARTMENT COUNT] departments generate [DATA VOLUME] of documentation, correspondence, and institutional knowledge. Retrieval is manual and inconsistent. Decisions are delayed while staff hunt for precedents, specifications, or historical context. Expertise walks out the door when people leave.

This is the bleed. It compounds weekly.

---

## Proposed Solution

A sovereign AI deployment by Oction Labs.

We install a private, retrieval-grounded AI system inside your infrastructure. Your documents are ingested into a per-tenant isolated knowledge base. Staff interact through a simple interface: ask a question, receive a cited answer grounded in your own materials.

**Key characteristics:**

- **On-premise.** Your data never leaves your network.
- **Cited answers.** Every response references its source document.
- **0% fabrication on trap questions** in our measured test set.
- **74% correct retrieval-grounded** vs ~5-10% without retrieval grounding.
- **Full audit trail.** Every query and response logged and timestamped.

---

## Scope of Work

### Phase 1: Discovery (Weeks 1-2)

- Stakeholder interviews to confirm use cases and success metrics
- Document inventory and source mapping
- Infrastructure assessment and hardware requirements
- Data classification and access control design

Deliverable: Discovery report with confirmed scope and technical specifications.

### Phase 2: Build (Weeks 3-8)

- System installation on agreed infrastructure
- Knowledge base architecture and document ingestion pipeline
- Retrieval engine configuration and accuracy tuning
- User interface deployment and authentication integration

Deliverable: Functional AI system ready for testing.

### Phase 3: Deploy (Weeks 9-10)

- User acceptance testing with sample queries
- Staff onboarding sessions
- Documentation and administrator training
- Go-live cutover

Deliverable: Production deployment with trained users.

### Phase 4: Operate (Ongoing)

- Monthly retainer covers: system monitoring, performance optimization, document re-ingestion, user support, and feature updates
- Quarterly business reviews to measure outcomes against baseline
- Continuous accuracy improvement and knowledge base expansion

---

## Timeline

| Phase | Duration | Typical Dates |
|-------|----------|---------------|
| Discovery | 2 weeks | Weeks 1-2 |
| Build | 6-8 weeks | Weeks 3-[TIMELINE] |
| Deploy | 2 weeks | Following Build |
| Operate | Ongoing | Monthly retainer |

Total implementation time: approximately 10-12 weeks from signature to production.

---

## Investment

**[TIER: Standard/Enterprise] Tier**

| Component | Amount (CAD) |
|-----------|--------------|
| Initial setup and deployment | $[TOTAL INVESTMENT] |
| Monthly retainer (ongoing) | $[MONTHLY RETAINER]/month |

*All amounts in Canadian dollars. Initial payment due on signature. Monthly retainer begins upon production deployment.*

### Available Add-ons

| Add-on | Fee |
|--------|-----|
| Additional department deployment | $15,000 |
| Legacy data digitization | $5,000 per TB |
| Expertise interview program | $2,000 per person |
| Custom integration development | $250 per hour |
| Standalone annual knowledge audit | $30,000 |

*Add-ons are optional and scoped separately if required.*

---

## Terms and Conditions

**Payment terms:**

- 50% deposit due on signature
- Balance due upon completion of deployment phase
- Monthly retainer invoiced monthly, due on receipt

**Cancellation:**

- Either party may terminate for material breach not cured within 15 days' written notice
- Monthly retainer may be cancelled with 30 days' notice
- Fees for work performed to termination date remain payable

**Confidentiality:**

- All client data handled per applicable privacy law (PIPEDA / GDPR as relevant)
- Oction maintains strict confidentiality of all proprietary information

**Liability:**

- Oction liability limited to re-performing affected work or refunding fees for that item
- Total liability not to exceed fees paid for affected services
- Client retains final review authority for all AI-generated outputs

**Intellectual property:**

- Client owns all ingested documents and generated outputs
- Oction retains pre-existing tools, frameworks, and system architecture
- Knowledge base and configuration remain client property

---

## Next Steps

1. **Review this proposal** with your decision-makers
2. **Schedule a discovery call** to confirm scope and address questions
3. **Execute agreement** upon mutual acceptance
4. **Begin Discovery phase** within 5 business days of signature

**Contact:**

Brandon Gill, CEO  
Oction Labs  
+1 (587) 896-4199  
brandon@oction.ai

---

*eight agents. one organism.*