Hours of work each week, across dozens of designers, now runs in minutes.

PRAGMA was built inside a 40+ year old family business. The system installed on PRAGMA clients was running here first — turning a weekly proof-deck pipeline that used to take hours of designer work into a Python run.

The 40-year backdrop

Brand 9 Signs was founded in 1986 out of a single van in Jacksonville, Florida. Four decades later it serves homebuilders, government agencies, military bases, universities, and Fortune 500 campuses across the country — design, fabrication, installation, and maintenance. The business has carried three generations of operating discipline and a customer base that doesn't churn. That kind of business doesn't get disrupted by a slide deck; it gets re-engineered from inside its own production loop.

Brand 9 Signs is a Hartley Capital company. PRAGMA — also a Hartley Capital company — was built around what happened next.

The problem (pre-DeckForge)

A national homebuilder engagement runs ~16 active communities at any given time. Each community produces a recurring brief: signage package, install schedule, brand spec, model identity, wayfinding, repair, offsite. Each brief produces a proof deck for client review — a designer pulling community data, applying the homebuilder's brand standards (specific typefaces, multiple wordmark variants, price-band overlays), assembling cover and interior pages in PowerPoint, exporting to PDF, sending for review, capturing revisions, cycling.

Multiplied across 16 communities, weekly: hours of designer work each week, distributed across dozens of designers. A six-figure annual labor envelope just to keep the proof-deck loop running — before a single sign is produced. This is the operational reality of every mid-market production-services business. It is also exactly the kind of work most companies look at and conclude: we'll just hire another designer.

What we built (DeckForge)

DeckForge is the internal automation pipeline. It is not a deck-design tool. It is the operator system that runs the entire sign-program loop end to end.

The loop:

• Brief. The Atlas dashboard's request column is the source of truth — what each community needs this week.
• Render. A Python engine consumes the brief and produces a brand-locked PPTX and PDF — covers, interior pages, sign instances, model identity, wayfinding, price overlays.
• Publish. An Apps Script web app receives the rendered files, saves them to the client Drive folder, sets sharing, and writes a row into the Atlas deck-status tab.
• Approve. Atlas's approval column moves through a lifecycle DeckForge writes itself: Drafting → Proof Sent → Revision Requested → Approved → Production-Rendered → Ordered → Installed.
• Produce. Once approved, DeckForge emits production press files (one row per physical sign location, with the correct CMYK profile, bleed, color spec) and dispatches to Signs365, local CNC, or local plotter.
• Install. Status flows back into Atlas through the existing install workflow — photos, monument signs, wayfinding, model ID, awning status.

What was hours of designer work each week, across dozens of designers, now happens in minutes.

The first major client's communities have already shipped through this pipeline: Grand Park, Marion Oaks, Bellehaven, Heath Preserve, Juliette Falls, Marion Ranch, Millwood, Pioneer Ranch, Grand Oaks Manor, Oak Hammock, Trailhead Landing, Tara Baywood, Tara Forest East, Tara Serena, Grand Park North, and the corporate office build — sixteen communities, one pipeline.

What 10x looks like

Adding a client at 10x scale — going from one national account to ten — is a brand/<client>/ folder commit. Drop in fonts, color tokens, wordmark variants, schema entries. The dashboard expands; DeckForge consumes it the same way. Zero new designers. The labor curve goes flat while the revenue curve goes up. Every additional client is pure margin against a fixed pipeline.

The Atlas dashboard built for the first national homebuilder has already helped Brand 9 land two additional national accounts. Both will get their own dashboard instances on the same architecture — same loop, different brand folder. "We ran ourselves at 10x without hiring" is the proof the next homebuilder, university campus, military base, or Fortune 500 retrofit client signs on.

What 100x looks like

At 100x — a thousand-plus communities, multi-vertical (residential, institutional, retail, hospitality, government) — the design bottleneck doesn't just shrink, it disappears. The constraint shifts entirely off the proof-deck loop and onto install-crew throughput, fabrication capacity, and brand-ingest velocity (how fast a new brand/<client>/ folder can be cut). Drones — currently in pilot for install supervision — turn install quality into a measured production line, not a hand-craft variable. At 100x the business looks like a nationally-scaled operator without the headcount of one. A Frontier Firm version of an old family business.

The PRAGMA implication

DeckForge is not a Brand 9 success story. It is the template.

Any business with repeatable client deliverables has a DeckForge waiting to be built — investor decks, legal briefs, sales collateral, sector reports, marketing kits, agency revisions, audit workpapers, compliance attestations, design handoffs, RFP responses. The pattern is the same: brief lives in a structured surface, agents render the deliverable in a brand-locked format, an approval lifecycle moves status forward, production output dispatches, and the loop closes.

This case study is not marketing. It is the working record — written by the firm that runs the system, about the operator the system was built on.