Case Study

The Intelligence Node: AI Lead Generation and a Second Brain for a Fractional Advisory Firm, Run Entirely In House

A New York fractional advisory firm wanted AI to run its prospect research and outreach drafting, but it could not let client data leave the building or let a machine send anything unsupervised. We designed an Intelligence Node that does both jobs: local models do the work, and a named person approves every send.

AI Automation · Delivered · By Nathaniel Rub, Founder · Published September 19, 2026

The client

AIM (AIManagement Inc.) worked with a fractional advisory firm founded in 2025 that serves clients on retainer across six practice areas, including capital strategy and AI and automation. The firm is not a broker dealer, an investment adviser or a placement agent, and every deliverable had to respect that line.

The challenge

The firm's outreach workflow ran six steps with 24 human handoffs, on top of a homegrown CRM holding roughly 8,500 contacts and 30,000 institutions. Institutional knowledge lived in inboxes, and earlier vendors had passed credentials around by email. Leadership wanted AI doing the legwork without sending a single client record to an outside model provider.

What we designed and built

The Intelligence Node has two connected halves, plus the controls that make them safe to run.

How it was delivered

An implementation plan with 20 architecture decision records, 48 tasks across phases zero through four, and a 20 item verification queue. It was written so an AI coding agent could execute it and a person could audit it line by line.

Where AI fit

Local models do the research and drafting, embeddings power retrieval across the vault, and scheduled routines keep the knowledge base current. Anything confidential stays on the firm's own hardware.

The baseline the build started from

MetricValue
Human handoffs per outreach cycle before the build24
Contacts in the homegrown CRMAbout 8,500
Institutions trackedAbout 30,000
Client data sent to outside AI providersNone, by design

Lead volume, research hours saved and the post build handoff count are not published, so no figures are stated here.

Why it matters for advisory firms

Advisory, legal and capital markets firms can rarely paste client data into public AI tools. Local models on ordinary business hardware are now fast enough for daily pipelines, so private AI is a practical default rather than a compromise.

FAQ

Why run the models locally instead of calling a cloud API?

Confidentiality and cost. Prospect and client data stays on the firm's hardware, and a daily pipeline on a local model has no per token bill.

What is a Second Brain in a business setting?

A searchable, linked knowledge base of everything the firm knows about its contacts, deals and decisions, kept current by AI rather than by memory.

Does this work with Microsoft 365?

Yes. This build was scoped for a Microsoft 365 environment, and the pipeline reads from and writes to whichever mailbox and CRM the firm already uses.

Related: AI Automation · Executive Dashboards · A Claude powered Gmail assistant for business development. See the rest of the work on the case studies index.

Work with AIM

Want AI doing the legwork without your client data leaving the firm?

Bring one workflow, its owner, and what an incorrect result would cost. That is enough to scope the first engagement.

Request a Consultation Agentic AI Solutions