Techallin
All case studies
Lore

How a 40-Person Advisory Firm Stopped Losing 3 Hours a Day to Document Search.

Meridian Advisory Group had 247 documents — HR policies, compliance manuals, research reports, client briefs — scattered across shared drives and inboxes. We deployed Lore and gave their entire team instant, cited answers from everything they'd ever written.

85%

Less Time Searching

From ~3 hrs/day to under 30 min

247

Documents Indexed

4.2 GB of institutional knowledge

<5s

Average Answer Time

Down from 45+ minutes manually

100%

Answers Cited

Every response links to source & page

Industry

Financial Advisory

Professional services · 40-person firm

Services

RAG Knowledge Base

Document indexing · Multi-tenant workspace

Timeline

6 Weeks

Discovery · Build · Indexing · Onboarding

Team

5 Specialists

1 Strategist · 1 AI Engineer · 1 Backend Dev · 1 Frontend Dev · 1 QA

Technology

React · Node.js

OpenAI Embeddings · Pinecone · PostgreSQL · AWS

What we were hired to solve.

Meridian Advisory Group operates across three practice areas — wealth management, compliance consulting, and M&A advisory. Over seven years, they had accumulated hundreds of internal documents: HR policies, ISO compliance frameworks, procurement guidelines, client onboarding procedures, and research reports.

The problem was invisible until someone calculated the cost. Every time a team member needed an answer — a specific policy clause, a compliance deadline, a client brief detail — they opened a shared drive, ran a keyword search, skimmed three wrong files, and eventually found the right paragraph twenty minutes later. Multiply that by 40 people, dozens of times a day.

They needed a system that could read everything they'd ever written and answer any question about it — instantly, accurately, and with a reference so anyone could verify the source.

The core problem: Their knowledge existed. It just wasn't accessible. Every answer was already written down somewhere — but finding it cost more time than the answer was worth. New hires took 3+ months to become independently useful. Senior consultants were answering the same questions repeatedly.

What we built.

A purpose-built system designed around the specific constraint — not a generic tool configured to fit.

RAG-Powered Document Retrieval

Consultants ask questions in plain language. Lore retrieves the semantically closest passages from indexed documents and generates a precise answer — with the source document, section name, and page number attached to every response. No hallucination. No guessing.

Document Upload & Instant Indexing

Drag and drop PDFs, DOCX, TXT, CSV, and PPTX files directly into Lore. The system chunks each document semantically and indexes it for retrieval — typically within 60 seconds. The moment a file is indexed, every member of the workspace can query it.

RAG Mode vs. Web Search Mode

Consultants toggle between two modes in a single click. RAG mode pulls answers exclusively from Meridian's indexed knowledge base — fully private, fully cited. Web Search mode lets them query the live internet for regulatory updates or market data. One tool, two contexts, zero switching cost.

Full Conversation History & Context

Every conversation is saved and searchable. Consultants pick up where they left off — no re-asking, no re-searching. The conversation list groups sessions by topic and document context, making it easy to revisit a compliance thread from last week or forward a cited answer to a colleague.

Built with production-grade tools.

No bolt-on integrations. Every tool chosen for the specific constraints of this project.

React
Node.js
PostgreSQL
AWS
OpenAI Embeddings
Pinecone (Vector DB)
RAG Pipeline
Multi-tenant Isolation

What changed.

Meridian's 12-person active team went from spending an estimated 3 hours per day searching documents to getting cited answers in under 5 seconds. Time previously lost to searching is now spent doing the work that actually matters — advising clients.

New analysts at Meridian used to spend their first three months learning where everything was. Now they ask Lore. From day one, a new hire can query the full scope of the firm's institutional knowledge — policies, procedures, compliance frameworks, client templates — and get an answer with a reference they can verify themselves.

The knowledge didn't change. The access to it did. Every answer Lore returns includes the source document, section, and page number. There are no incorrect citations because every citation is a direct retrieval — not a summary, not a paraphrase.

Before

  • 3+ hours/day lost to document searches
  • Inconsistent answers across the team
  • New hires took 3 months to work independently
  • Senior staff answering the same questions on repeat
  • No single source of truth

After

  • Answers in under 5 seconds, cited to the exact page
  • Consistent responses across all 12 members
  • New hires productive from week one
  • 247 documents instantly queryable
  • Knowledge compounds instead of hiding
The moment we indexed our compliance library, I asked Lore a question I'd been answering manually for three years. It gave me the answer, the section number, and the exact page. I forwarded it to a client in 30 seconds. That used to take me 40 minutes.

Sarah Okafor

Head of Compliance · Meridian Advisory Group

Common
questions

Tell us what you're building.We'll tell you how to solve it.

No pitch. No pressure. Tell our AI about your challenge — it will match you with the most relevant solution and show you exactly how we'd approach it.