Schematic of a decision scorecard: two options weighed row by row with a recommendation band

Greystone Capital

A boutique investment advisory firm asked for an honest, no-vendor-interest read on where AI belongs in compliance work — and where it absolutely does not.

Greystone CapitalOne system
  • Priorities scoped to real…
  • One explicit no-go
  • Roadmap → internal build
Shared data · role-based access3 of the modules shown
Schematic of the shipped system, drawn from the case notes

The challenge

Before

Greystone Capital is a boutique investment advisory firm with around 40 employees. Their compliance team was drowning in document routing and quarterly client reporting — the kind of repetitive judgement work that looks like a perfect AI use case until you read the regulatory rules. Leadership had been approached by half a dozen "AI for finance" SaaS vendors, each with a confident pitch and zero appreciation for the risk surface. The directors wanted a written, vendor-neutral read on where AI belonged in the operation — and, just as importantly, where it didn't.

  • Compliance lead manually routing client documents across 4 reviewer queues
  • Quarterly client reports drafted from scratch each cycle by ops team
  • KYC pre-screen handled manually with no documented decision criteria
  • Six "AI for finance" vendor pitches active simultaneously with no evaluation framework
  • Leadership unsure which workflows carried regulatory risk if automated
  • No internal benchmarks for hours-saved-per-workflow
  • Previous SaaS trial abandoned mid-pilot because nobody had scoped the use case
  • No written roadmap to take to the board for AI investment approval
BeforeDisconnected tools
  • Compliance lead manually routing client…
  • Quarterly client reports drafted from…
  • KYC pre-screen handled manually with no…
  • Six "AI for finance" vendor pitches active…
  • Leadership unsure which workflows carried…
  • No internal benchmarks for…
Schematic of the starting point, drawn from the case notes

The solution

What we built

We ran our 1-week paid AI Operations Audit. We interviewed the compliance lead, the ops director, and the head of client services. We walked the document-routing pipeline end-to-end, sat through a quarterly reporting cycle, and mapped the KYC pre-screen in detail. We scored 18 workflows on hours saved, automation complexity, and regulatory risk. The deliverable was a written report identifying three priority candidates: document classification (high leverage, low risk — green light), quarterly report drafting (medium leverage, medium risk, human-in-the-loop required), and a third candidate flagged in red — KYC pre-screen. We explicitly recommended against automating KYC without a human reviewer on every decision, with the regulatory reasoning written out. The client built doc classification internally using our roadmap as the brief.

How the system flows

  1. Discovery interviewscompliance lead, ops director, head of client services
  2. Workflow mapping18 ops and compliance workflows scoped end-to-end
  3. Risk scoringregulatory exposure weighted alongside hours saved
  4. Document routing analysis4 reviewer queues mapped with decision criteria
  5. Quarterly reporting cycle walkthroughbottleneck identified at draft stage
  6. KYC pre-screen reviewflagged red, explicit no-go with regulatory reasoning
  7. Priced roadmaptop 3 workflows scoped against real engagement budgets
  8. Written reportPDF + Notion, formatted for board-level review
Schematic of a decision scorecard: two options weighed row by row with a recommendation band
Schematic of a decision scorecard: two options weighed row by row with a recommendation band

Need an honest read on AI risk in a regulated business?

Our 1-week AI Operations Audit is vendor-neutral on purpose. We'll tell you which workflows AI should touch, and which ones it absolutely should not — in writing.

No retainer lock-in · Month-to-month · Full transparency