Zero Fog
Reference Case · 03 of 04

The right idea, to the right desk, at the right moment.

The Situation: A U.S.-headquartered institutional brokerage with a strong equity research franchise was losing the timing battle on the buy side. Research output was high-quality but reached client desks fragmented, delayed, and unmatched to the actual mandates each fund was managing. The constraint was not insight quality. It was insight delivery.
The Constraint We Identified

The product was research. The actual value was relevance at velocity.

Institutional clients receive research from dozens of sources. They have minutes — often seconds — to act on what's relevant. A research note that arrives well-written but undifferentiated from the rest of the morning's inbound goes unread. A research note that arrives matched to a fund's current mandate, with a clear read on why this matters to this portfolio, gets the trade.

The firm was producing the first kind of note at scale. The second kind required pre-call work that analysts and salespeople didn't have the time or the data to do consistently. The day was being spent assembling context that should have been structurally available. The actual research time was being squeezed by the assembly time.

What We Built

A structured mirror of fund holdings and mandates, with applied AI synthesis layered on top.

Structural pattern Insight-to-client flow
RAW SIGNALS → APPLIED SYNTHESIS → CLIENT-MATCHED INSIGHT SIGNALS Equity research Filings & transcripts Holdings & mandates Prior engagement APPLIED AI SYNTHESIS Domain-tuned, not generic Theme & signal extraction Mandate-to-insight match Relevance prioritization Suppression of noise CLIENT OUTREACH Hedge fund A3 idea matches Mutual fund B1 thematic alert Pension manager CSuppressed (noise) RESULT · MANUAL OVERHEAD − 60% · BUY-SIDE RATING + 15% YoY Same research output. Different signal hitting the right desk at the right moment.
The structural pattern. The research itself didn't change. What changed was the layer between research and outreach — a structured mirror of who held what, a domain-tuned synthesis engine on top, and suppression logic that protected client attention from undifferentiated noise.
  • MirrorWe built a unified knowledge graph of funds, holdings, mandates, prior engagement history, and analyst output. The mirror made structurally available what the salesforce had been assembling by hand on every call.
  • CapabilitiesThe relevant capability dimension was clearly Velocity — the firm's ability to move from signal to client-matched outreach at speed. The constraint was at the synthesis layer between the research desk and the salesperson's day.
  • AnalyzeWe engineered a domain-specific AI synthesis layer — explicitly not generic GPT prompting, which the firm had already tried and found unreliable for the depth required. The synthesis layer extracted themes from research, matched them to fund mandates, prioritized relevance, and suppressed noise.
  • ExecuteWorkflow automation routed the synthesized output through the firm's existing CRM into structured outreach. Quarterly cycles tuned the relevance models against real client feedback. The salesforce's day moved from assembly to conversation.
What the Engagement Produced

Same research, sharper signal.

−60%
Manual synthesis overhead
Time the salesforce spent assembling context before client conversations.
+15% YoY
Buy-side ratings
The institutional buy-side surveys that drive commission allocation. Multiple drivers contributed.
Higher
Trading-decision influence
Measured through engagement velocity at the moment funds were allocating. Qualitatively confirmed by senior client feedback.

The 60% reduction is the cleanest read of the structural change — it's measuring the work that was directly removed from the salesforce's day. The 15% buy-side rating shift is meaningful but noisier; better research quality from other parts of the firm also contributed in the same period.

If you're moving signal through human assembly when it should be moving through structure, the diagnostic is the first step.

30 minutes · Senior practitioner · No deck

Request fit call
Other reference cases
01 · Broker-dealer execution

From fragmented operations to a unified advisor experience.
02 · Post-acquisition operating model

Two operating models, one synergy thesis — and a fog condition between them.
04 · Strategic intelligence velocity

1.5 months → 1.5 minutes — without losing depth.