Investor Report

1. Executive Summary

ProPlum Network has crossed two critical thresholds this week. First, the direct sales engine is live: Samantha 2, our AI-powered callback dialer, deployed September 17 on Cloudflare's global edge. Every lead is manually dialed by a human rep, every answered call runs through a conversational booking flow ending in a calendar-committed callback, and every voicemail drops automatically.

Second, the affiliate-channel question — our highest-leverage distribution bet — is now answered with empirical evidence, not opinion. A 500-agent AI market simulation (25 agents × 10 rounds × 2 controlled variants — an internal exercise run at the founder's own expense, no company capital touched) produced a clear verdict on pricing, trust mechanics, and objection handling. It validated a staged go-to-market strategy that spends under $500 before requesting any network-scale capital.

Recommendation in one line: Fund the direct engine's first 90 days and a five-pilot affiliate cohort from our current operating budget; defer the 25K–50K network raise until Gate 1 produces proven, recurring cash flow.

2. Where the Business Stands

The direct channel economics: Sam subscriptions range from $199 to $999/month, with voice usage billed per second. We restrict access to one partner per market. A single closer utilizing Samantha can work 40–50 leads per morning. There is no revenue share and zero Customer Acquisition Cost (CAC) beyond time.

3. The Affiliate Question

The strategic question put to the simulation: Can a third-party creator channel sell Sam — and on what terms?

About the Simulation Engine — MiroFish

The evidence in this section was produced by MiroFish, an open-source multi-agent swarm-intelligence engine designed to predict trends and social contagion by seeding thousands of interactive LLM agents into parallel digital worlds — in the project's own words, "SimCity meets AI forecasting."

"Traditional prediction models treat the world like a math equation. You feed in numbers, you get numbers out. But the real world doesn't work that way... MiroFish simulates the messy, social dynamics of the real world using thousands of AI agents that talk, argue, persuade, and evolve — just like people do."

For this study, ProPlum rehearsed its affiliate launch inside two parallel 25-agent worlds and let the dynamics run. The rehearsal was an out-of-pocket expense borne personally by our founder. What follows is what the sandbox taught us before a single real dollar — or a single real creator — was spent.

Methodology

Two parallel swarms of 25 agents each (20 micro-creators, 1.8k–23k followers; 5 macro creators, 60k–210k) lived through a 10-round recruiting narrative: outreach, peer friction, market reality, then a divergence point at round 7.

• Variant A (Written Terms): The vendor responds with terms in writing — a signed 24-month rate-lock agreement, a proof-before-promotion package (raw call recordings, latency logs, churn cohorts, two callable reference plumbers), bundled minutes, written metro exclusivity, and an automatic tier-up path.
• Variant B (Silence): The vendor never puts terms in writing.

Scale: 365,000 tokens of simulated deliberation, zero dropped agents — an internal out-of-pocket expense borne by our founder.

Findings

1. Trust is the binding constraint — and it is solvable for the cost of a signature. The single largest friction cluster in the simulation was "comp not in writing" (249 objection records across 25 of 25 agents). Without paper, 28% went to a hard "no." The round the signed rate-lock landed, the cohort's average stance jumped +49 points to +67.5. In Variant A, 25 of 25 agents signed at the standard offer. The percentage split was never the obstacle; the signature was.

2. Bundled minutes are the only sellable pricing shape through this channel. In the control arm, agents independently wrote bundled minutes into their counteroffers unprompted. With bundles on the table (e.g., 400 min/$499, single consolidated invoice), the per-minute pricing objection went effectively extinct. Pass-through usage billing (our direct-channel model) is unsellable through affiliates. Verdict: Bundle-first at every tier for channel-differentiated packaging.

3. Objections are finite, ranked, and answerable. Seven distinct objection patterns surfaced, each with counter-messages that measurably flipped holdouts: comp-in-writing, proof-of-product, churn transparency, brand risk, payout timing, exclusivity scope, and FTC exposure.

4. Macro creators cost more, and only at scale. Cohort B counters settled at 38–50% recurring. That is the price of macro distribution — rational only when the direct channel is saturated and reference inventory is deep.

Decision Matrix (Willing to pitch, by effective Year-1 $ per close)

Gate 0 verdict: PASS. Both stated pass conditions cleared. We now have a predictable, objection-proof playbook for creator onboarding.

4. Pro Forma: 90-Day Operating & Gate 1 Pilot Budget

To justify funding Stage 1 and Stage 2 from the current operating budget, we have modeled the cash impact of the next 90 days. The model assumes conservative direct-channel conversion and isolates the minimal capital required to prove the affiliate channel.

Direct Channel (Samantha 2) — 90-Day Unit Economics & Projections

Target: Convert 25 of the 192+ active leads at an average blended tier of $300/mo.

Gate 1 Affiliate Pilot — Hard Costs

Target: 5 hand-picked micro-creators.

Conclusion: The direct engine achieves profitability almost instantly due to zero CAC. This easily subsidizes the $500 required to test the simulation's affiliate findings in the real world.

5. The Honest Caveats

An LLM-agent simulation measures direction, not destiny. Human creators are messier than simulated ones — they stall, ghost, and improvise. That is precisely why we gate capital behind human proof: the simulation tells us what to bring (paper, proof, bundles); five real creators in 60 days will tell us who shows up.

Furthermore, the simulation tested the affiliate channel in isolation, not head-to-head against direct CAC. The direct channel keeps 100% margin and is live today. The affiliate channel trades 20–50% recurring margin for outsourced hustle. Sequencing these correctly is vital.

6. Recommendation & Capital Plan

Stage 1 — Direct engine, now through Day 90 (Operating Budget). Run Samantha 2's first batches against the 192-lead inventory. Target: 20–50 closed plumbers to establish a churn baseline and secure reference recordings. This produces the exact proof package the simulation dictates we need to flip affiliates.

Stage 2 — Gate 1 affiliate pilots (<$500 from operating cash flow). Recruit five hand-picked micro-creators from the GHL/agency-builder community. Each receives the signed 24-month rate-lock, proof package, bundled-minutes pricing, and a direct line to our closer.

Pass condition: ≥2 closes in 60 days and ≥1 affiliate self-reporting the pitch as easy.

Stage 3 — Network scale (Deferred, triggered by Gate 1 evidence). Only after real closes will we build the affiliate portal, payout ledger, and Tier-2 builder white-label. The previously modeled $25K (proof of channel conversion) and $50K (the compounding machine) remain the right-sized asks — but the simulation proved we should buy our initial validation with $500 of pilots, not $25K of infrastructure.

7. Risk Register

8. The Ask

We are asking for zero external capital today.

Watch two metrics over the next 60 days: direct closes from Samantha batches and Gate 1 pilot closes. Once those numbers validate the pro forma above, the $25K ask converts from speculative risk to pure acceleration fuel. The simulation proves that the room we walk into with paper, proof, and bundles is a room that signs. We are spending the next 60 days printing that proof.

Prepared by the ProPlum Network strategy office. Sources: internal feasibility dossier (§5A Market Simulation, MiroFish v2, run Sep 17 2026), live account audits, deployed-system verification. Items to verify with counsel before external distribution: FTC penalty figures; final rate-lock agreement language.

ProPlum Network — Confidential.