Know everything your projects know.

Turn scattered documents, decisions, and history into a living knowledge base. AI works as your project memory, not a search box.

🔒 Runs in your Azure tenant 🔗 Syncs with SharePoint & Teams ⚡ Every answer traced to source
"We have five years of project documents across SharePoint, email, and a dozen shared drives. When our lead engineer left, the reasoning behind half our key decisions left with him, and nobody could explain why we'd made certain calls."
faster than manual
document search
100%
answers traced to
source document
0
knowledge lost when
people leave
The Solution

A living memory layer for your projects.

This isn't a folder search, and it isn't a static wiki. It's a structured reasoning system that understands your project history the way your most experienced team member would, connecting decisions, documents, and conversations across time, format, and source.

Ask a question in plain English and get an answer with the source document, the reasoning, and links to related decisions. There's no migration and no manual tagging: the system connects directly to SharePoint, Teams, and email. Your team decides, and the system makes sure nothing gets forgotten.

Natural-language queries Permission-aware Full traceability Azure-native Human decides

Architecture Overview

📁

Your Project Knowledge

Documents, emails, Teams messages, and SharePoint sites, connected directly with no migration or manual tagging required

🧠

Knowledge Layer

Azure AI Search · Azure AI Foundry · artifact graph with hybrid vector + keyword retrieval

⚙️

Reasoning Engine

Proprietary orchestration framework for multi-agent, loop-based reasoning with goal completion

📤

Delivery

Azure Functions · Azure AD permissions · Word / Excel export · Teams-native, structured UI in your tenant

Core Capabilities

Four capabilities. One project archive.

Every capability runs on the same underlying knowledge layer. Start with one project, then expand to your entire archive without rebuilding the foundation.

🎯

Natural-language project search

Ask questions about any project in plain English, with no more guessing which folder, who to ask, or what date something happened. The system searches across SharePoint, email, and Teams to find the answer, not just a matching file.

Natural-language queries Cross-source search Temporal understanding No migration required
🧩

Reads context, not just keywords

The strongest answers rarely live in a single document. The system understands that a decision made in a meeting note connects to the email thread that followed and the contract that resulted, and reasons across all of them.

Multi-document reasoning Relationship mapping Temporal linking Cross-format understanding
🔐

Permission-aware by design

Every answer respects your existing SharePoint and Azure AD permissions, enforced at query time. A restricted document never surfaces its content to someone without access: the system tells you it exists, without showing you what it says.

Per-document permissions Query-time enforcement Azure AD native No data leakage
💬

Natural-language project Q&A

Multi-agent sessions with per-user history mean the system remembers context across a conversation. Ask "who worked on the Meridian project?" then "what changed in the contract last quarter?" and every answer cites its source.

Multi-agent sessions Source-traced answers Goal-completion reasoning Full transparency into steps
The Interface

Built for project teams,
not engineers.

A rich review interface that replaces the black-box answer and the folder listing. Every answer is filterable, drillable, and exportable. Built for project teams and stakeholders who need to trust and verify what they're told.

  • Browse the full knowledge graph or search by project, person, or date
  • Click any answer to see the source document and reasoning chain
  • Compare related decisions or documents side by side
  • Human review on every answer, so your team stays in control
  • One-click export to Word, Excel, or share directly in Teams
Knowledge Base: Meridian Project (1,240 documents · 6 data sources)
Query: "Why was the Q3 deadline extended?" 3 sources found · full reasoning chain
ANSWERED Deadline extended by 6 weeks due to vendor delay
Project status meeting notes (Sept 12) confirm the extension was approved after the primary vendor reported a 6-week component delay. Confirmed in a follow-up email from the Project Lead to stakeholders the same week.
Meeting_Notes_Sept12.docx + Email thread · Confidence: High
REVIEW Conflicting budget figures across two documents
Budget_Tracker.xlsx shows €420,000 remaining; Q3_Report.pptx shows €380,000. The documents were updated on different dates, and Budget_Tracker is more recent (Oct 3 vs Sept 20).
Cross-document comparison · Budget_Tracker.xlsx + Q3_Report.pptx
NO ACCESS Contract_Amendment_Final.pdf (restricted)
This document matches your query but is restricted to the Finance team. Permission is enforced at query time, so no content was surfaced.
Access controlled by your Azure AD / SharePoint permissions
Getting Started

Running on a real project within one week

There's no migration and no manual tagging. The system connects to your Azure environment and works with the SharePoint sites, email, and Teams channels you already have.

1

Discovery workshop

We map your project knowledge sources, key repositories, and search pain points. You leave with a concrete pilot scope and proposal.

½ DAY · FIXED FEE
2

Pilot on a live project

We deploy the system on a real project's document set: your actual sources, your actual structure. Typically 2 to 3 weeks.

2–3 WEEKS
3

First live queries

Your team runs the first real queries on a production project. We support the review session and tune the system together.

DAY 1 OF PRODUCTION
4

Expand & evolve

Add projects, additional data sources, or a Teams-native rollout to the wider organisation on the same foundation.

OPTIONAL RETAINER
Where Keyword Search Fails

The connection a folder search misses, found with evidence.

Keyword search finds the document with the right words in it. It doesn't find the email thread that explains why a decision was made, or the meeting note that shows a contract was later amended. The system reasons across formats, time, and sources, so the answer reflects what actually happened rather than just which file matches your search terms.

Meeting notes ↔ Emails Decisions ↔ Outcomes Contracts ↔ Amendments + full project history

Live example: connecting the dots

Question: "Why did the budget increase in Q3?"
Meeting note: "Approved additional €40k for vendor delay" (Sept 12)
Email confirms: Finance approved the transfer the same week
Result: ANSWERED: with the full decision chain, not just a budget line

Every answer shows the connected sources and the reasoning chain behind it.

Where It's Used

Any workflow where project history needs to be found and trusted

The system adapts to your sources and structure, not the other way around.

📚

Project onboarding

New team members get instant, evidence-backed answers instead of waiting on tribal knowledge held by whoever has been there longest.

🔍

Historical decision lookup

"Why did we choose X?" answered in seconds, with the original reasoning and the exact document or email that captured it.

🗂️

Cross-project pattern mining

Find how similar problems were solved on past projects: proposal reuse, risk patterns, vendor history, without relying on memory.

🤝

Client & stakeholder handover

Hand over a project with full context intact, not just a folder of files nobody has time to read before the next milestone.

⚖️

Compliance & audit readiness

Trace any decision back to its source, with a full reasoning trail ready for internal review or an external audit request.

🔄

Knowledge continuity

When someone leaves, their institutional knowledge doesn't leave with them. The reasoning behind past decisions stays searchable.

Why It Matters

Real impact on project continuity and trust

Faster than manual document search

What used to take a project manager an afternoon of digging through folders and asking colleagues now takes seconds, with every answer sourced.

100%
Source-traced answers

Every answer cites the exact document, email, or meeting note that produced it. No guessing, no invented history.

Projects covered, one knowledge layer

Add new projects and data sources without rebuilding the knowledge base from scratch. It grows as your organisation does.

Common Questions

Straight answers

Does this replace our project management tools?+
No, it sits alongside them. Your PM tools (Jira, Azure DevOps, Planner) track tasks, status, and workflow. This system captures the reasoning, history, and context behind decisions: the "why" rather than the "what's the status." Most teams use both: the PM tool for execution, this for institutional memory.
How does it handle permissions across different sources?+
Every source keeps its own access rules. The system reads your existing SharePoint and Azure AD permissions and enforces them at query time, so a person only sees answers built from documents they're already allowed to access. If a restricted document is relevant, the system indicates it exists without exposing its content.
Will it give wrong answers about our project history?+
Every answer is grounded in your actual documents and cites its source. The system does not invent decisions or fill gaps with plausible-sounding guesses. Where evidence is ambiguous or conflicting, it says so explicitly, as shown in the "REVIEW" status in the interface. Think of it as a reasoning layer over your own project archive, rather than a general-purpose chatbot with opinions.
Our projects generate new documents constantly. Does it stay current?+
Yes. The system connects to your live SharePoint, email, and Teams sources rather than a static export, so newly created or updated documents are reflected without manual re-indexing or reconfiguration.
Our project data is highly sensitive. How is it protected?+
Everything runs within your Azure tenant. Documents are processed by your Azure AI Foundry deployment, and no data is sent to external APIs. Your existing Azure AD access controls and data residency settings apply from day one. We provide a Data Processing Agreement for your procurement and legal review.
How long does a typical deployment take?+
A pilot on a live project with your real document set typically takes 2–3 weeks from kickoff. This gives you a working system on your actual sources, with your actual structure, not a demo built to look good in a pitch. Full production deployment, if you choose to proceed, typically follows within 4–6 additional weeks depending on the number of sources involved.
Ready to see it working?

See it running on your projects.
Real documents, real answers.

The best way to evaluate this system is a live demo on a real document set from one of your actual projects, anonymised if needed. Book a 30-minute call, and we'll walk through a live query, then tell you honestly whether your use case is a fit.

No commitment · No slide deck · Engineers talking through your problem · Based in Switzerland, operating across EU & UK