AI Knowledge Assistant - PwC

The Challenge

When consultants opened new NARM consultations, understanding how similar situations had been handled previously often required finding the right historical consultation or speaking with experienced specialists. Much of that institutional knowledge was difficult to access across closed engagements.

This created three problems:

• Historical knowledge was difficult to surface.
• Routine guidance depended heavily on senior experts.
• Sensitive client and consultation information needed to remain protected.

Objectives

• Surface relevant precedent on demand.
• Reduce dependency on senior SMEs for routine guidance.
• Make historical consultations searchable.
• Protect confidential information.

What Athen Built

Athen built an AI Knowledge Assistant that allows consultants to ask questions in natural language and retrieve relevant guidance from past NARM consultations. The solution combines conversational retrieval, secure knowledge indexing, anonymisation and permission-aware access.

How It Works

Searchable Knowledge Base

Closed consultations are securely ingested, indexed and made retrievable.

Conversational Retrieval

Consultants ask questions naturally rather than manually searching through historical cases.

Privacy-Preserving AI

Company names, individuals and client details are anonymised before information is presented.

Permission-Aware Access

Users only retrieve information aligned with their access and compliance policies.

Business Impact

• Faster access to institutional knowledge
• Reduced routine workload for senior experts
• More consistent guidance across similar scenarios
• Protected confidential information

Technology

Azure OpenAI · RAG · Anonymisation · RBAC

Why It Matters

For professional services organisations, valuable knowledge often exists—but remains difficult to access. Athen turns that institutional knowledge into a secure, searchable business capability.