Consulting Firm Intelligence

Ask about past client work. Answers cite the exact document and page or section. Below is a recorded example: the question, the answer, the documents it retrieved and the execution pipeline.

ContextAll clients (8)See the search filters ›

Filters are shown for reference. Log in to choose clients, industries, engagements and documents.

Add clients, engagements and documents in Clients. Search always runs with your own permissions.

Example question · All clients (8)

What did we recommend on shrink, and what did the pilot achieve?

This is a read-only public demo. Log in to ask your own questions.

Public research

ModelLog in to use

Example response · not a live AI executionCompletedgroq/openai/gpt-oss-120b5054 ms

Recorded Nov 10, 2026, 7:27 a.m. and verified by a partner. Open the cited documents below to check each reference.

- Recommendation on shrink: Implement a shrink‑reduction programme that targets the 22 high‑loss stores by tightening fresh‑department ordering discipline, improving markdown timing, and adding targeted loss‑prevention staffingD1. - Pilot achievement: In a 12‑store pilot (January – June 2025) the programme cut shrink from 2.9 % of sales to 2.2 % of salesD1. The evidence does not provide results for the broader rollout beyond the pilot.

›Client documents1Northwind Retail

[D1] Store Margin Recovery Final ReportOpen page 3
Northwind Retail · Store margin recovery · Final report
MERIDIAN ADVISORY · CONFIDENTIAL · SYNTHETIC DEMONSTRATION DOCUMENT — FICTIONAL CLIENT 3. Recommendations and pilot results Promotion ROI gating. Every promotion above 20% depth requires a pre-event ROI estimate and post-event review. Promotions failing two consecutive reviews are discontinued. Shrink reduction program…

Tools used: Client documents. Click a reference such as [D1] to open its source.

DEMO REPLAY

Execution pipeline

5054 ms

Replaying the recorded trace of a verified run (recorded Nov 10, 2026, 7:27 a.m.) — no AI call is being made.

✓
INPUTUser query received
0 ms
mode: Live + automatic fallbacktask: answeragent: ATLAS Assistantsolution: consultingcharacters: 64
✓
SECURITYPermissions & access checks
1295 ms
- ✓Registered in AI inventory— atlas-assistant
- ✓Active— active
- ✓Role permitted— partner ∈ {partner, consultant}
- ✓Client wall (database RLS)— 8 client(s) accessible
✓
AGENTAssistant plan: tools selected by rules
0 ms
task: answermodel: defaultfallback: yessolution: Consulting Firm Intelligence
- ✓Client documents— Consulting questions are answered from client documents.
✓
SUPABASESupabase: client knowledge search
1880 ms
mode: hybridclients: 8function: match_chunks_scopereturned: 8security: RLS as the signed-in userdocuments: allengagements: allclients matched: 5embedding model: gemini/gemini-embedding-001
✓
SOURCESKnowledge sources
0 ms
passages: 8documents: 7
- ·[D1] Northwind Retail — Store Margin Recovery Final Report p.3— Final report · confidential · rrf 0.0323
- ·[D2] Cedar Health Network — Emergency Department Flow Final Report p.1— Final report · confidential · rrf 0.0308
- ·[D3] Northwind Retail — Store Margin Recovery Final Report p.2— Final report · confidential · rrf 0.0304
- ·[D4] Aurora Telecom — Churn Reduction Programme Final Report p.2— Final report · confidential · rrf 0.0301
- ·[D5] Pinecrest Insurance — Underwriting Productivity Final Report p.1— Final report · confidential · rrf 0.0299
- ·[D6] Northwind Retail — Inventory Shrink Diagnostic Proposal p.1— Proposal · confidential · rrf 0.0297
- ·[D7] Keystone Manufacturing — Predictive Maintenance Pilot Proposal p.1— Proposal · confidential · rrf 0.0297
- ·[D8] Keystone Manufacturing — Plant Operations Benchmark Final Report p.2— Final report · confidential · rrf 0.0291
✓
MODELModel invocation
1745 ms
mode: automatic fallbackmodel: gpt-oss 120B (Groq)billing: freeattempts: 1provider: groqrequested: defaultinput tokens: 2107open weights: yesfallback used: nooutput tokens: 369
- ✓#1 groq/openai/gpt-oss-120b— 1342 ms · 2107→369 tokens
✓
GENERATEAnswer generation
0 ms
characters: 465finish reason: stopoutput tokens: 369reasoning shown: no
✓
VALIDATEEvidence, numbers & governance validation
1 ms
- ✓Every reference points to real evidence— 1 cited
- ✓Every figure appears in the evidence— ok
- ✓Answer is grounded— cites its evidence
- ✓Evidence fenced as data— no tag-like text found
- ✓Data boundary— 8 client(s) in scope, RLS-filtered
✓
OUTPUTFinal response
0 ms
Completed
total ms: 5054

Sign in to Meridian Advisory

Consulting Firm Intelligence

Sign in to Meridian Advisory

Execution pipeline

Sign in to Meridian Advisory

Consulting Firm Intelligence

Sign in to Meridian Advisory

Execution pipeline