Keel
Structured intelligence for your firm's data and documents.
Every financial firm accumulates years of data, documents, and institutional knowledge. Most of it ends up unstructured, unsearchable, and underused, scattered across file shares, practice management systems, email, and cloud storage.
Foundry is Keel's proprietary engine for making sense of it all. It starts with the hardest layer, your documents, and turns them into structured, searchable, governed assets that land in SharePoint and the rest of your Microsoft environment ready to use.
The Problem
Engagement letters, financial statements, client correspondence, internal memos, compliance records. Thousands of files accumulating every year, each one a node in the firm's institutional knowledge.
But when it's time to migrate, consolidate, or simply find something, firms discover that none of it is organized. Metadata is missing or inconsistent. File names are meaningless. Duplicates are everywhere. Critical documents sit next to drafts and junk with no way to tell them apart.
The result: migrations stall, platforms underperform, and institutional knowledge stays locked away.
When Firms Need Foundry
The firm is moving to a new platform or consolidating systems and discovers their data and documents are a mess. Metadata is missing, naming is inconsistent, and no one knows what's critical and what's junk. Foundry inventories and classifies everything before anything moves.
The platform is in place but documents were dumped in without structure. Search doesn't work because there's nothing meaningful to search on. Foundry brings order to what's already there by extracting real metadata and applying classification so the platform actually performs.
Institutional knowledge is locked in filenames and folder hierarchies that only a few people understand. When those people leave, the knowledge leaves with them. Foundry extracts metadata from the documents themselves, turning tribal knowledge into structured, searchable data.
Two firms' worth of data, documents, naming conventions, and folder structures need to become one. Nobody wants to do it manually, and nobody trusts a bulk move. Foundry normalizes across systems, classifying, deduplicating, and delivering a unified document set ready for a single platform.
How Foundry Works
Foundry processes your firm's documents through four stages, each building on the last.
We start by connecting to wherever your data and documents live. File servers, SharePoint, cloud storage, practice management systems, email archives. Foundry maps the full landscape before anything moves.
You can't organize what you haven't inventoried.
Foundry pulls from every source and stages documents into a single, unified view. Metadata, content signals, and structural information are extracted from every document in its native format, creating a complete inventory with a full audit trail of what was found, where it came from, and what it contains.
For many firms, the majority of documents can be classified from native metadata alone, before any AI processing is needed.
For documents that need deeper analysis, Foundry layers intelligence progressively. Rules-based matching handles known patterns. Lightweight AI classification handles the next tier. Only genuinely ambiguous cases get full AI processing.
The output: every document classified, tagged, and enriched with structured metadata.
Clean, structured, classified data delivered into your platform. SharePoint is the primary destination for most firms, with metadata, classification, and governance landing alongside the documents themselves so the platform is intelligent from day one. Whether you're migrating to a new system, consolidating into SharePoint, or organizing what you already have, Foundry delivers documents that are findable, governed, and useful from the moment they arrive.
What Makes Foundry Different
Most tools treat every document as a black box. Foundry reads the native structure of every major format (.docx, .pdf, .xlsx, .pptx, .msg, .eml), extracting metadata that's already there before any AI processing is needed.
Native extraction first. Rules-based classification second. Lightweight AI third. Full processing only when necessary. Each layer validates and enriches the next.
Foundry understands the language of financial services. It recognizes document types, naming conventions, and structural patterns specific to accounting, advisory, and compliance work. This domain knowledge is built in, not bolted on.
Foundry doesn't just move documents into new folders. It produces structured metadata (document type, client, entity, engagement, confidence score) that makes your platform intelligent from the start.
