Case study · Black Mountain Financial

How a one-person lending brokerage built its own AI infrastructure in 90 days — one particle at a time.

Black Mountain Financial is a boutique commercial and residential lending brokerage in Canberra — a single principal, its own credit licence, and an average deal size of $2–3M. In ninety days Quanta built a working AI stack across the practice: CRM automation, research monitoring, content production, document generation — without a single big-bang project and without ever stopping client work to do it.

Client
Black Mountain Financial
Sector
Lending brokerage · ACL holder
Size
Single principal
Build time
90 days · operational
Founding build

George Popadalis is the principal of both Black Mountain Financial and Quanta Consulting. Quanta built this stack on George's own regulated practice before offering the methodology to outside clients. What you're reading is the build log of a business that runs on it every day.

The situation

A regulated practice with the same problem every small operator has.

Black Mountain Financial (BMF) is a boutique commercial and residential lending brokerage in Canberra, run by a single principal under its own credit licence. Deals span business borrowing, property development, and residential lending. The work is high-value, complex, and entirely on one set of shoulders.

What it didn't have was a back office. Every credit memo, every broker submission, every BID compliance note, every entity structure chart, every email reply was being produced by hand. Document templates lived in folders. The CRM held data but didn't actually do anything. Research on competitor lenders happened when there was time — which there never was.

In one line

“I'm a one-person practice handling $2–3M deals with the operational discipline of a sole trader. Something has to change but I can't stop client work to fix it.”

This is the exact problem most professional services firms are stuck inside — high-quality, high-margin work being delivered through low-quality, manual systems, with no time to fix it because the work itself is consuming every hour. Quanta was built specifically to solve this problem for practices like BMF: small enough to move fast, focused enough to stay in the practice's lane.

The specific problems

The approach

No transformation programme. One particle at a time.

The wrong answer would have been an “AI transformation” — a multi-month engagement, a thirty-page strategy document, a system bought before anyone knew what they needed. Quanta's answer was the opposite: identify the single most expensive piece of manual work each fortnight, build one small thing that fixes it, ship it into the practice, and let the results decide what comes next.

Every particle followed the same rules:

What follows is the actual sequence of particles that shipped at BMF inside ninety days. Each one is real. Each one is in production. Cumulatively they form the operating stack the practice runs on today.

The particles

Ninety days. Twelve particles. One working stack.

What follows is the actual chronology — what shipped, in what order, and what each piece does for the practice today.

Particle 01Foundations · Week 1

The brand system.

Before anything else, BMF needed to look like itself. Quanta built the brand system first — colour palette (Mountain Black, Peacock Green, Warm Ivory), typography pairing (Bebas Neue, IBM Plex Serif, Outfit), a written voice guide, and three reusable document templates — and locked it before any AI work began.

What shipped
  • Full brand specification document, hosted
  • Reusable Word templates for credit memos, broker notes, BID commentary
  • Email signature block, landing page, business card templates
  • Tone-of-voice guide for client comms
Particle 02CRM & data · Weeks 2–3

The CRM bridge.

The broker CRM held the data but was effectively read-only from outside the platform. Quanta built a secure data bridge that exposed client, deal, asset, liability, and party data on demand to the practice's internal AI tools — within Australia, on SOC 2-certified infrastructure.

The effect: client and deal context could now be pulled into any document, email, or analysis in seconds, without rekeying. The CRM stopped being a passive database and became an active part of the work.

What shipped
  • Secure data bridge between CRM and internal tools
  • Hosted in Australia on certified infrastructure
  • Used daily for credit memos, BID notes, broker submissions
Particle 03Compliance · Week 4

The BID notes skill.

Best Interests Duty compliance — the regulatory requirement for mortgage brokers to document why a recommended product is in the client's best interest — was being written by hand for every file, with constant risk of inconsistency. Quanta built a reusable skill that pulls the CRM opportunity data, applies the compliant three-section structure, and produces a draft that needs only review, not authorship.

What shipped
  • Structured BID notes skill (customer objectives, lender rationale, suitability)
  • Pre-fills from the CRM automatically
  • Outputs branded Word document ready for the file
Particle 04Compliance · Week 5

The broker submission notes skill.

The companion piece to BID. Where BID is client-facing, broker submission notes are the narrative that accompanies the application to the lender's credit team. Quanta used the same pattern: pull the data from the CRM, apply the structured format the lender expects, output a branded document ready to send.

What shipped
  • Lender-facing submission notes skill
  • Variants tuned for different lenders' formats
  • Reduced submission preparation time from hours to minutes
Particle 05Market intelligence · Week 6

The research monitoring agent.

Lender policy changes, regulatory updates, and competitor product launches were being picked up by accident. Quanta built a monitoring agent that scans the relevant sources daily and surfaces anything that matters to the practice — running autonomously on secured Australian infrastructure, with notifications delivered through BMF's existing comms stack.

What shipped
  • Autonomous daily research agent on certified AU hosting
  • Structured tracking of policy changes and competitor moves
  • Notifications routed through the practice's existing tools
  • Dashboard for reviewing accumulated intelligence over time
Particle 06Inbox · Week 7

The inbox triage agent.

Lender notifications, broker queries, automated bank emails, and admin pings were arriving at hundreds per week. Quanta added an inbox triage agent inside BMF's existing email stack — it classifies incoming mail, surfaces what needs attention, and quietly files the rest. Time-to-zero on the inbox dropped from “never” to “before lunch.”

What shipped
  • Inbox triage agent operating within the existing email platform
  • Smart de-duplication so nothing important gets buried
  • Daily digest of what was actioned and what needs a human
Particle 07Content infrastructure · Week 8

The content engine.

Quanta set up a long-form article and LinkedIn publishing program: a content brief skill for development finance, a separate skill for business finance, a Notion-based pipeline for tracking drafts, and the practice tone-of-voice guide enforced at the skill level. Brief-to-published lead time dropped from days to hours.

What shipped
  • Two persona-specific content skills (devfin, bizfin)
  • Notion content management system with editorial workflow
  • Four-pillar 12-month writing program
  • First seven long-form articles produced and ready to publish
Particle 08Client comms · Week 9

The email engine.

Quanta built five reusable client email templates — welcome, settlement, rate review, referrer thank-you, decline — as a skill that produces on-brand HTML emails with the BMF design system applied, Outlook fallback included. Every client touchpoint now looks like BMF, automatically.

What shipped
  • Five branded HTML email templates + skill to generate them
  • Outlook-compatible fallbacks tested
  • Variable substitution from CRM data
Particle 09Delivery · Week 10

The entity group structure tool.

Lending submissions for commercial deals require a visual entity structure diagram — who owns what, how the borrowers connect, where the guarantors sit. These were being hand-drawn. Quanta built a reusable structure-chart skill that takes the entity list and produces an interactive, on-brand HTML diagram ready for credit submissions.

What shipped
  • Entity structure skill producing branded HTML diagrams
  • Variants for borrowers, guarantors, trusts, related parties
  • Now used for every commercial submission
Particle 10Memory · Week 11

The practice memory layer.

Working across multiple clients, deals, and projects meant context kept getting reset. Quanta built a persistent memory layer so that practice-wide context — deals in flight, client preferences, structural decisions — persists across sessions and is queryable on demand. Hosted in Australia, encrypted, and owned by the practice.

What shipped
  • Persistent memory service on certified AU hosting
  • Structured store of practice-wide context
  • Queryable on demand by internal AI tools
Particle 11Web · Week 12

The landing page system.

One-page deal microsites for specific submissions and co-branded partner pages were being mocked up in Word. Quanta built a proper site infrastructure that lets the practice spin up a branded, mobile-responsive page in hours, hosted under the practice domain on certified AU infrastructure.

What shipped
  • Reusable site infrastructure with brand baked in
  • Page templates that produce a new microsite in hours
  • First co-branded partner microsite live
Particle 12Modelling · Week 13

The financial modelling library.

Each complex deal previously meant building a financial model from scratch. Quanta assembled and standardised a library of reusable Excel models — three-way cash flow templates, development feasibility, commercial serviceability, lender comparison. Model-to-submission time dropped dramatically.

What shipped
  • Reusable three-way cash flow template
  • Development feasibility model (multi-stage)
  • Serviceability and lender comparison templates
  • All branded, all version-controlled

Twelve particles. One working stack.

None of these were built as a single transformation programme. Each one shipped on its own, in its own month, and started delivering value the day after it went live. Together, they're the operating system the practice runs on today — built without ever stopping client work to do it.

The outcome

What changed in the practice.

The honest answer: small things, compounded over ninety days, in a way that's hard to see week-to-week but very visible at the end of a quarter.

~60%
Document time
Reduction in time-to-first-draft on credit memos, broker notes, and BID commentary. From a half-day to under an hour for typical files.
90 days
No big-bang projects
Ninety days of continuous incremental shipping. No transformation programmes. No multi-week pauses. No external dev team.
$0
Software licensing per seat
The whole stack runs on the practice's own infrastructure and owned IP — hosted in Australia, on SOC 2-certified servers. No vendor lock-in, no per-seat SaaS fees.

More important than any of the numbers: the practice runs on its own infrastructure now, and the principal has the headspace back. When a new opportunity comes in, the question is no longer “where do I find time?” — it's “which particle would help here?”

What we learned

Five things this build taught us about how small operators should actually adopt AI.

These lessons came directly out of the BMF build and are now the operating principles Quanta applies to every client.

A note on provenance.

George Popadalis is the principal of both Black Mountain Financial and Quanta Consulting. The work here was not a client engagement — it was Quanta building on its own practice, which is why every particle is in production and none of it is theoretical. We think that's a stronger credential than a reference from a client who was asked to trust it first.

If you'd prefer to see independent client work before engaging, that's a fair ask. Third-party case studies will be published from mid-2026.

Could your practice run like that?

Quanta builds these stacks for other small operators using the exact same particle-by-particle method. Start with a 30-minute conversation — no charge, no slides, no commitment.

Book a free 30-minute call