Maik Business Plan

Working name: Maik. A note on the name: throughout this plan I use the notation "M-" rather than committing to the working name "Maik." During brand validation in Phase 10 we surfaced a prior-art conflict with MAiK Lifestyle (Edinburgh, UK), which markets soft furnishings into adjacent design-conscious channels. Final name resolution sits in the Phase 10 brand sprint and will be locked before public launch. For internal planning purposes assume continuity of strategy regardless of which two-syllable mark we land on.

Prepared by Steve Aylward, MD, Ven Agency. 16 June 2026.

1. Executive Summary

M- is an Australian-made, AI-operated, premium 3D-printed sculptural lighting brand. We design objects whose forms are only possible because AI generates the geometry and farm-scale 3D printers manufacture them. Every piece ships from a Melbourne light-industrial studio, lit by USB-C and built in batches small enough that no two pieces are exactly alike.

What M- is: a single brand, in a single category, at launch. A hybrid catalogue of 12 to 20 evergreen sculptural lamps priced $129 to $499 AUD, punctuated by four numbered drops per year of 50 to 200 units each. A direct-to-consumer Shopify Plus store run by a 15-agent fleet (Forge orchestrator plus 14 specialists), backed by Ven Agency's full stack of creative, paid media, SEO, EDM, and operations capability.

What M- is not: a marketplace shop, an Etsy seller, a commodity 3D-print store, a multi-category lifestyle brand, a B2B-first business, or a wholesale-first business. It is also not a labour-intensive studio. The whole point of the experiment is to prove that a brand of this calibre can be operated by an AI fleet with one human owner in the loop on brand-critical decisions.

The unfair advantage is threefold. First, Ven's full stack sits behind M- from day one: paid media operators, SEO compounding, EDM craft, design system, analytics pipeline, and the agent-orchestration substrate (Multica) that lets a 15-agent fleet coordinate without falling over. Second, the bet on generative AI for industrial design plus desktop print farm economics arrives at exactly the moment both technologies cross a price-performance threshold. Third, Melbourne supplies a design-literate domestic audience large enough to seed and a credible AU-made narrative that imported competitors cannot match.

The ask is $150,000 AUD initial capital, fully committed by Ven, plus a $100,000 milestone-gated tranche releasing at the 6-month checkpoint subject to KPIs in Section 11.

Expected outcomes. 12 months: $250-400k cumulative revenue, design-press recognition in at least two of Yellowtrace, Broadsheet, Habitus Living, Vault, or Inside, agent fleet operating at 70 percent or higher autonomy in graduated workflows, second printer tranche purchased. 24 months: $600k-1.2M annualised revenue, 15+ printers, a defensible "voice-adapter" IP layer for parametric customisation, optional second sub-brand, and a Ven case study published to compound credibility for the agency itself.

2. Positioning

M- sits at the intersection of three movements that have until now been treated as separate: premium 3D-printed lighting, generative-AI industrial design, and AI-operated D2C commerce. Each has a credible reference point. None has been combined deliberately.

Substrate. The substrate is PLA and PETG bioplastic, with selective use of woodfill and recycled-content filaments for textural variation. Substrate restraint is core to the brand: we publish the material, the printer model, the print time, and the wattage of each piece, the way Aesop publishes the botanical lineage of every formulation.

Category. Sculptural domestic lighting in the $129 to $499 AOV band. Table lamps and pendants at launch, floor lamps from Q3 (M-004). We do not sell ambient strip lighting, smart bulbs, ceiling fixtures requiring AC certification, or commodity desk lamps.

Direct competitive landscape.

Gantri (US, Los Angeles) is the obvious reference. Premium 3D-printed lighting, designer collaborations, $300-$900 USD, AC-certified. Their structural lesson: a print-on-demand catalogue can sustain premium pricing if the geometry justifies it. Where they leave room: their AC-electrical model gates them on speed-to-market for new SKUs and limits them to one production facility. M- operates USB-C-only at launch, which collapses certification overhead.
Wooj Design (Brooklyn) prove that a brand can scale on Instagram-driven D2C alone with desktop printers. They are the operational template. Where they leave room: their geometry is signature but human-designed; we add the generative layer.
Creme Atelier (Helsinki) and Sonogo (NL) confirm that the European market accepts $200-$500 EUR for printed lighting in design-press channels. They validate price elasticity at the upper end.
Aerend, Lumio, Pablo sit adjacent in premium domestic lighting without 3D-print. They define the visual quality bar customers will judge us against on the shelf.

Structural references (not direct competitors).

Aesop: substrate-led restraint, archival product naming (M-001, M-002), refusal to discount, retail theatre.
MUJI: catalogue depth without category sprawl, the discipline of saying no to product extensions.
Nilufar Gallery (Milan): collector-grade limited drops with serialised provenance. The drop cadence borrows from here directly.

What M- deliberately is not. Not a maker collective. Not a one-off commissioned studio (though we accept commissions as a tertiary revenue stream). Not a marketplace tenant. Not a smart-lighting brand. Not multi-category. Not a sustainability-first brand that leads with bioplastic credentials, although we will substantiate them. The brand leads with form.

3. Ideal Customer Profile

Primary persona: the design-conscious AU urban renter-to-owner, age 28 to 50.

Household income $120,000 AUD or higher. Lives in inner Melbourne, Sydney, Brisbane, Perth, or Adelaide. Rents an apartment of 60-90 sqm or owns a first home of 90-140 sqm. Cares about objects rather than furniture, in the sense that they will spend $400 on a single lamp before they spend $4,000 on a sofa, because the lamp is portable, expressive, and signals taste with lower commitment.

Information diet: Pinterest (mood-boarding their space), Instagram (saved posts of Yellowtrace, Est Living, The Local Project), TikTok (design and home tours), Habitus Living, Vault, Inside, Broadsheet on weekends, occasionally Dezeen for international reference. Listens to Local Project podcast, reads Monocle in print.

Buying triggers: a new apartment, a relationship moving to cohabitation, a renovation completing, a birthday or housewarming gift for a similarly design-literate friend, a single discovery moment on Instagram or in a Yellowtrace feature.

Objections: AU-made premium pricing, lead time uncertainty, "is this just a 3D-printed novelty," fragility concerns, plug-pack confusion (the USB-C question must be answered cleanly on PDP).

Secondary persona: the gift buyer for a design-conscious recipient.

Slightly older, 35-55, less design-literate themselves but buying for a partner, sibling, or close friend who is. AOV skews higher because they trade certainty for taste signalling. They need gift-wrap, gift-message capture at checkout, and a confidence-inspiring brand site. Klaviyo gifting flow targets Mother's Day, Father's Day, end-of-financial-year, and December.

Tertiary persona (trade): the AU interior designer or stylist specifying for residential and small hospitality projects. Trade portal access at 30-40 percent off retail, project NET 30 terms after first paid project, specification PDFs available on request.

4. Value Proposition

Three pillars, each defensible.

Pillar one: AI-designed generative geometry. Every M- form starts in a generative pipeline trained on a curated reference set of sculptural lighting, organic structures, and architectural ornament. We do not claim AI replaces design; we claim AI enables forms that human CAD workflows are too slow to explore. The output is then human-curated, refined, and proofed by Steve and the Ven design team before it enters the catalogue. The marketing claim is honest: forms that are only possible because AI designs them and 3D printers make them.

Pillar two: AU-made on demand. Every piece ships from Melbourne, printed to order or in small batched runs. Lead time honesty is core: PDPs state "ships in 5-7 business days" rather than the dishonest "in stock" of overseas dropshippers. The sustainability claim is substantiated: bioplastic substrate, no shipping container, no warehouse holding inventory in dead stock, recyclable packaging via noissue. We do not lead with the green claim, but we do not hide it either.

Pillar three: no two pieces exactly alike. Print-to-print variance is a feature, not a defect. Layer line variation, micro-warping in cooling, parametric seed variation across runs (especially in drops) all contribute to a serialised uniqueness story that 3D-print at scale can offer and injection-moulded competitors cannot. Each drop unit is numbered and signed digitally. Provenance card ships with every piece.

Why a buyer pays $189-$499 instead of $89 for a Kmart equivalent. They are not buying a lamp. They are buying a sculptural object that lights, with a story that is legible on Instagram, an origin that is verifiable on the brand site (factory livestream, named printer, print log), and a level of design intention that the commodity tier cannot fake.

5. Pricing Strategy

Launch range $129 to $499 AUD. Anchor SKUs:

M-001 Table lamp — $189. The hero. Volume driver. Lives in the catalogue indefinitely.
M-002 Pendant — $229. Second-volume SKU. Catalogue evergreen.
M-003 Drop 1 — $329, numbered 1/100. Limited to 100 pieces. Sold once, then archived.
M-004 Floor lamp — $429, releasing Q3 of year one. Catalogue evergreen.
M-005 Drop 2 — $399 to $499, Q4 of year one, 50 to 75 pieces.

Drop pricing premium 20 to 40 percent over evergreen of equivalent footprint. Drops are numbered, signed digitally, archived after sellout. We never re-release a drop.

Bundling is not encouraged. A second-lamp incentive would dilute the per-object reverence we are pricing into the catalogue. Where customers want two pieces, they pay for two pieces.

Free AU shipping over $189 covers the psychological barrier at the M-001 price point. Below $189 (if we ever introduce a sub-anchor accessory SKU) shipping is calculated at checkout. International shipping is paid by customer, calculated, and explicitly flagged as 10-14 day lead time on top of production lead time.

No discounting. No promo codes. No EOFY sale. No Black Friday. The only price-down mechanism is trade pricing (Section 6) and gallery consignment (Section 6). This is non-negotiable in the brand position and matches Aesop's discipline.

6. Revenue Model

Primary: D2C Shopify Plus. Target mix 75 percent year one, 70 percent year two, 65 percent year three. The primacy is intentional. D2C protects margin (no wholesale haircut), preserves brand control over presentation and copy, and feeds the first-party data pipeline (Klaviyo, GA4, Meta pixel) that the agent fleet needs to optimise paid and EDM.

Secondary: Trade and wholesale to interior designers and architects. 10 percent year one ramping to 20 percent year three. Trade portal launches month 6 with 30-40 percent off retail, NET 30 after first paid project, project specification PDFs on demand. Targets the AU specifier community, especially Melbourne and Sydney studios that already feature in Yellowtrace and The Local Project. Trade volume per account is small but compounding: a single specifier placing two lamps into one residential project per quarter, across 20 accounts, generates meaningful and predictable revenue.

Tertiary: Gallery consignment. Two to three Melbourne and Sydney design galleries on consignment at the start. Targets: Modern Times Melbourne, Criteria Collection Melbourne, Stylecraft showrooms (Melbourne and Sydney). 5 to 10 percent of revenue year one, ramping to 15 percent year three. Galleries take 40-50 percent. The trade-off is margin for brand legitimacy. Being shown alongside Pierre Jeanneret reproductions and Australian ceramic studios elevates positioning faster than any paid media tactic.

Quaternary: Custom commissions. Parametric variations of catalogue SKUs sold direct: pick a base SKU, request a colour or seed variation, pay $750 to $2,500. 5 percent year one ramping to 15 percent year three. The commission stream also captures B2B small-hospitality work (a Melbourne wine bar wanting six matched pendants tuned to its colour palette, for example).

Explicitly rejected. Etsy AU and other marketplace channels. Off-brand, race-to-bottom on price, no customer data ownership, no brand control. Amazon AU equally rejected. Any future marketplace conversation has to clear a brand-erosion hurdle that we do not believe it will.

7. Go-to-Market Plan

Three waves across year one.

Wave 1 — Months 1 to 3: launch.

The launch hook is a single hero lamp (M-001) plus M-003 drop 1, executed with three coordinated assets:

A 24/7 transparent factory livestream from the Melbourne studio (single camera, ambient sound, OctoEverywhere feed embedded into the brand site as a footer ribbon). The livestream is the brand's anti-marketing marketing. It says: this is where it is made, watch us make it.
A Yellowtrace exclusive cover story, pitched 8 weeks ahead of launch, embargoed to launch day. Yellowtrace gets first access to the M-001 hero image set, a written piece on the AI generative pipeline, and an interview with Steve.
A coordinated PR push across Broadsheet, Habitus Living, Vault, Inside, and The Design Files. Each gets a tailored angle (Broadsheet: Melbourne maker story; Habitus: design-led production; Vault: collectible drop angle; Inside: substrate and sustainability; Design Files: the founder narrative).

Paid spend $10k per month in Wave 1: Meta and Google split roughly 70/30, Meta-weighted because the visual product favours feed and Reels. Pinterest spend deferred to Wave 2 until we have organic Pin engagement data to optimise against.

Klaviyo welcome flow live at launch: a 5-email sequence introducing brand, factory, substrate, drop cadence, and the M-001 PDP. Q4 drop teaser sequence begins month 2.

Organic compounding: Pulse (social agent) seeds Instagram and TikTok daily, Sage (SEO agent) begins long-tail SEO content production at 2-3 articles per week from month 1.

Wave 2 — Months 4 to 9: compound.

SEO content engine hits compounding velocity. Sage produces 2-3 long-form pieces per week on topics like "Australian-made designer lighting under $300," "best 3D printed pendant lamps 2026," "how to choose a sculptural floor lamp for a rental apartment," "the rise of generative design in Australian product design." Long-tail target keyword set of 200+ queries, each scored by Sage against ranking difficulty and commercial intent.

Trade outreach begins month 4: direct introduction email sequence (Smith agent drafts, Steve approves and sends per org rule) to a curated list of 80 Melbourne and Sydney interior design studios. Trade portal launches month 6.

Second drop (M-005) launches end of Q3 / start of Q4. Pre-announced 4 weeks ahead via EDM (Ember agent). Numbered, signed, single-purchase per customer. Expected to sell out in 24-72 hours based on Wave 1 EDM engagement signals.

Paid scales to $12-15k per month from month 5 if ROAS holds above 2.0x.

Wave 3 — Months 10 to 12: harvest and plan.

Year-1 retrospective campaign across owned channels. Press push for end-of-year design coverage (Yellowtrace year-in-review, Broadsheet best-of, Inside annual). Year-2 planning sprint: do we add a second category, do we spin a sub-brand, do we open a second print bay. The data answers the question, not founder gut.

8. Operating Model

Print farm. 8x BambuLab P1S plus 1x X1C, deployed in a 120 sqm light-industrial studio in inner Melbourne (Brunswick, Collingwood, or Cremorne North preferred). The X1C is reserved for prototyping, larger-format hero pieces, and the materials that require chambered printing. The P1S fleet handles production volume.

Studio fit-out requirements:

Dedicated 32A electrical circuit, RCD-protected, sub-board metered for per-printer power tracking.
Bofa AD350 fume extraction with HEPA + activated carbon, ducted to a shared manifold across the print bay.
Sealed humidity-controlled filament storage at 30 percent relative humidity, with 25kg bulk hoppers and per-spool desiccant for active queue.
Monitoring via OctoEverywhere across the full fleet, with Anvil (fulfilment scheduling agent) integrated to the queue API.
24/7 camera for the factory livestream, fixed position, ambient lighting on a slow loop.
Climate control: split-system air conditioning, target 22-24°C ambient.
Lease target: 18-month with 18-month extension option. $4,500 to $6,500 per month gross including outgoings.

Atelier operations. Anvil schedules the print queue. Inputs: open orders, promised dispatch dates, printer availability, filament stock, drop-deadline overlays. Output: a rolling 5-day print schedule with manual override panel for Steve. Re-runs every hour. Hearth (inventory + supplier reorder) reads Anvil output and triggers filament reorders against the dual-source supplier set (X3D AU and Aurarum AU) at the 25kg/month volume threshold for discount.

Fulfilment. In-house pick, pack, dispatch up to 500 orders per month. At 500 orders per month we hand over to a Sydney 3PL (eStore Logistics or James and James). The handover trigger is volume, not month. Packaging custom-printed via noissue or Packhelp, MOQ 500, brand-controlled colour and finish.

Supplier strategy.

Hardware: BambuLab AU plus Prusa AU as dual hardware vendors by month 4. Prusa MK4S serves as the failover platform if BambuLab AU support deteriorates or BambuLab firmware policy changes restrict the fleet.
Filament: X3D and Aurarum dual-source from day one. Volume discount triggered at 25kg per month per vendor.
Electronics: Jaycar and Altronics for AU-certified components on prototypes. Shenzhen OEM for scale on USB-C ports, cables, and any future internal componentry, with RCM and SAA documentation for any AC future.
Packaging: noissue (recycled, custom-print) primary, Packhelp secondary, MOQ 500.

Quality control. Every unit photographed before dispatch (Lumen agent automated capture station with consistent lighting), filed against order ID. 4 percent reprint cost baked into margin model from day one. Visible-defect threshold documented and applied by the dispatch operator (the only human-only QC gate in the workflow).

Electrical certification path. USB-C only at launch. Customer brings their own USB-C PD plug pack, or uses an existing one. This decision eliminates RCM and SAA certification overhead, eliminates plug-pack supply chain risk, and signals an Apple-grade restraint that the audience will read as deliberate. Future AC variants (specifically the floor lamp range from year two) gated by certified OEM plug-pack with documented compliance, sourced via a Shenzhen OEM with RCM and SAA paperwork verified before any AC SKU enters the catalogue.

9. Team Plan — The Agent Fleet as the Team

The team is Forge (orchestrator) plus 14 named specialist agents. Steve is the sole human in the loop on brand-critical and outbound-comms work for the first 90 days. Multica sits across the fleet as meta-observability and audit log.

Autonomy ramps in three steps: month 3, month 6, month 12. At each ramp, Steve reviews the prior 30 days of agent decisions, marks safe-to-automate and require-approval, and Forge re-codes the boundary in agent config.

Forge — orchestrator. Coordinates the fleet, allocates tasks, resolves conflicts, surfaces approval-queue items to Steve. Autonomy: full within the agreed envelope, escalates anything novel. Reports to Steve via the Monday digest (compiled by Oracle).

Atlas — listings. Owns the Shopify product catalogue. Creates and edits PDPs, manages product taxonomy, ensures consistency of substrate / dimensions / weight / lead time fields across SKUs. Autonomy month 1: propose-and-approve on all new listings. Month 3: autonomous on edits to existing listings, propose-and-approve on new. Month 6: autonomous on listings, Steve reviews drop SKUs only. MCP: Shopify.

Lumen — photography. Coordinates the QC photography station (every unit photographed before dispatch) and the catalogue / lifestyle shoot pipeline. Brief generation, shot list, post-production direction. Autonomy: autonomous on QC capture, propose-and-approve on catalogue and campaign shoots. MCP: Shopify, Google Drive, Lightroom integration via local file watcher.

Quill — copy. Owns brand copy across PDPs, EDMs, landing pages, paid creative. Voice-locked to a brand guide. Autonomy month 1: propose-and-approve on all surfaces. Month 3: autonomous on PDP edits and EDM body, propose-and-approve on hero pages and paid headlines. Month 6: autonomous on all copy except drop launch and PR. MCP: Shopify, Klaviyo, Google Drive.

Beacon — paid media. Meta Ads and Google Ads campaign build, monitoring, optimisation. Autonomy month 1: propose-and-approve on all campaign launches, autonomous on within-campaign budget shifts under $200/day. Month 3: autonomous on campaign optimisation, propose-and-approve on new campaign launches and creative rotation. Month 6: autonomous on optimisation and rotation, Steve reviews quarterly. MCP: Meta Ads, Google Ads, GA4, Shopify.

Sage — SEO. Long-form content production, on-page optimisation, GSC monitoring, technical SEO. Autonomy: autonomous on content production and publishing within a published topic plan. Propose-and-approve on topic plan changes and any pages that touch brand-positioning copy. MCP: Shopify, Google Search Console, DataForSEO, GA4.

Ember — EDM. Klaviyo flows, campaigns, segmentation, deliverability. Autonomy month 1: propose-and-approve on all sends. Month 3: autonomous on flow optimisation and segment sends under 5k recipients, propose-and-approve on campaigns to full list. Month 6: autonomous on campaigns to full list, Steve reviews drop sends only. MCP: Klaviyo, Shopify.

Anvil — fulfilment scheduling. Reads orders, schedules the print queue across the 9-printer fleet, prioritises by promised dispatch date, flags backlogs. Autonomy: autonomous within the queue logic. Escalates to Steve only on missed-SLA risk. MCP: Shopify, OctoEverywhere, internal queue store.

Hearth — inventory and supplier reorder. Reads Anvil's queue, tracks filament stock against forecast usage, triggers reorder against dual-source suppliers at threshold. Autonomy month 1: propose-and-approve on all purchase orders. Month 3: autonomous on reorder against agreed supplier and price ceiling. Month 6: autonomous on all reorder, Steve reviews quarterly cost trend. MCP: Shopify, supplier portals (X3D, Aurarum), internal inventory store.

Smith — supplier comms. Drafts all outbound supplier email and partner communication. Per the organisation rule, Smith never sends emails without Steve approval. This is a hard human-only gate, indefinitely. Smith drafts, surfaces to Steve, Steve sends. MCP: Gmail (draft only).

Echo — customer service triage. First-touch on customer enquiries via help@ inbox and Shopify Inbox. Routes to FAQ-answer auto-reply where the question is clearly answered (lead time, materials, sizing, gift options). Escalates to Steve for anything that touches refunds, quality complaints, drop allocation, or media enquiries. Autonomy month 1: propose-and-approve on all replies. Month 3: autonomous on FAQ-tier replies, propose-and-approve on anything else. Month 6: autonomous on FAQ and shipping-status, propose-and-approve on quality and refund tickets. MCP: Gmail, Shopify Inbox, Help Scout if added later.

Mercury — returns and refunds. Process refund and replacement workflow. Reads Shopify orders, validates return reason against policy, issues refund or replacement print order. Autonomy month 1: propose-and-approve on all refunds. Month 3: autonomous on refunds under $250 with photo-validated defect, propose-and-approve above. Month 6: autonomous on all defect refunds, Steve reviews monthly aggregate. MCP: Shopify, Anvil queue.

Pulse — social. Instagram, TikTok, Pinterest organic content. Caption, schedule, light editing. Autonomy month 1: propose-and-approve on all posts. Month 3: autonomous on scheduled grid posts and Stories, propose-and-approve on Reels and TikTok originals. Month 6: autonomous on grid and Stories, propose-and-approve on Reels with brand-critical hooks. MCP: Instagram Graph API, TikTok Business, Pinterest Business.

Oracle — analytics. GA4, Shopify Analytics, Meta and Google Ads reporting, custom dashboard build. Compiles the Monday digest for Steve, every Monday at 7am AEST. Autonomy: fully autonomous on reporting and analysis. Cannot take action; only surfaces. MCP: GA4, Shopify, Meta Ads, Google Ads, Klaviyo, Google Search Console.

Herald — drop coordination. Owns drop launch timeline. Coordinates Quill, Lumen, Ember, Pulse, Beacon, and Anvil into a coordinated drop sequence. Autonomy: propose-and-approve indefinitely. Drop launches are brand-critical and remain Steve-gated. MCP: cross-fleet calendar, Klaviyo, Shopify.

Multica sits across all of the above as the meta-observability layer: agent decisions, approval queue depth, escalations, audit log, project allocation. Every new body of work in M- gets a Multica project before agent tasks are queued against it.

Human gates summary. Steve approves: all outbound email Smith drafts (indefinite), all drop launches (indefinite for first year), all PR responses, all refund decisions above $250 in the first 6 months, all new SKU launches, all changes to the brand voice guide.

10. 3-Year Financial Summary

Detailed numbers are the Financial Modeller agent's deliverable. This section sets the assumptions and shape.

Revenue model assumptions.

AOV blended $189-$229 across the catalogue. Drops skew higher ($329-$499) but are a small share of unit volume.
D2C 75 percent year one, 70 percent year two, 65 percent year three. The shift down reflects deliberate diversification into trade, gallery, and commission as the brand earns specifier-level legitimacy.
Cumulative revenue targets: Y1 $250-400k, Y2 $600k-1.1M, Y3 $1.2M-2.4M. The wide band reflects honest uncertainty on paid efficiency in months 6-12 and on drop sell-through.
Customer repeat rate target: 15 percent in year one, 25 percent year two, 30 percent year three. Repeat behaviour in this category is slow but compounding, and drops are the main repeat-triggering mechanism.

Key cost categories.

Capex year one $85k. Hardware $30k (printers, filament storage, fume extraction, electrical fit-out, monitoring), studio setup and bond $15k, brand and identity $15k, legal and IP $5k, paid media seed $10k, working capital and contingency $10k.
Opex month 12 $30k/month. Studio rent $5-6k, filament and consumables $4-6k, paid media $10-15k, software stack (Shopify Plus, Klaviyo, Multica, monitoring, design tools) $2-3k, fulfilment supplies and shipping $2-3k, agent fleet compute and API costs $1-2k, contingency $1-2k.
Opex month 24 $55k/month. Studio scale (second bay or expanded floor) $7-9k, filament $8-12k at higher throughput, paid media $20-25k, fulfilment outsourced to 3PL $5-8k at volume, software stack $3-4k, agent compute $2-3k.

EBITDA shape.

Year one: negative. Burn is expected and budgeted. Capex front-loaded, revenue ramps over 12 months, marketing investment compounds.
Year two: breakeven Q3. Compounded SEO, repeat purchase, drop reliability, trade revenue all hit critical mass in months 18-21.
Year three: profitable. Net margin target 15-22 percent depending on revenue band hit. Margin upside comes from the agent-fleet operating leverage (one human, fixed at Steve's overhead allocation, rather than scaling team headcount with revenue).

Capital ask. $150k initial committed by Ven at month zero. $100k tranche released at month 6 subject to the milestone KPIs in Section 11.

11. Milestones

90 days.

Store live on Shopify Plus, M-001 and M-002 listed.
Hero drop M-003 launched and sold through 50 units minimum.
First $50k cumulative revenue.
Factory livestream live and embedded on the brand site.
Klaviyo welcome flow live, list at 750+ subscribers.
Yellowtrace exclusive published.
Agent fleet operational across all 15 roles with autonomy at month-1 settings.
Multica observability layer running, weekly Oracle digest delivered.

6 months.

$100-180k cumulative revenue.
Second drop M-005 announced, EDM teaser sequence in market.
EDM list at 3,000+ subscribers.
GA4, Meta pixel, Google Ads conversion tracking validated clean by Tracking Monitor.
Yellowtrace plus one other publication feature published.
Trade portal launched, first 10 trade accounts approved.
Agent fleet autonomy at month-6 settings, approval-queue depth below 20 items at any time.
$100k milestone tranche released subject to all of the above.

12 months.

$250-400k cumulative revenue.
Design press recognition across at least two of Yellowtrace, Broadsheet, Habitus, Vault, Inside, The Design Files.
Year-two expansion decision documented: second category yes/no, sub-brand yes/no, second print bay yes/no.
Agent fleet operating at 70 percent autonomy across graduated workflows.
Second printer tranche purchased (4-6 additional P1S units, taking fleet to 13-15 printers).
Gallery consignment live with at least one Melbourne and one Sydney gallery.
Repeat purchase rate at 15 percent or higher.

24 months.

$600k-1.2M annualised revenue.
1 to 3 brands under operation (the second and third only if earned by year-one and year-two performance, not assumed).
15+ printers in operation.
Voice-adapter IP layer defensible: parametric customisation pipeline filed as a process patent or trade secret with documented protection.
White-label B2B option live or rejected with documented rationale.
Ven case study published, M- referenced as proof of the AI-operated brand thesis.
Fulfilment handed to 3PL, in-house dispatch retired.

12. KPI Dashboard Structure

Oracle compiles the Monday digest for Steve every Monday at 7am AEST. The digest is a single dashboard view with six top-level metrics, one operational health block, and a flagged exceptions section.

Top-level (six).

Revenue — rolling 28-day and cumulative-to-date, with prior-period comparison and run-rate to next milestone.
Gross margin — by SKU and blended, with substrate cost, reprint cost, and shipping subsidy stripped out so the number is honest.
Blended CAC — paid spend plus content allocation divided by new customer count. Tracked against AOV to maintain LTV-to-CAC discipline.
Repeat rate — 90-day and 180-day repeat purchase rate, with cohort view from acquisition month.
Print farm utilisation — printer-hours-active versus printer-hours-available across the fleet. Target 65-80 percent. Above 80 percent means we need more capacity; below 50 percent means we are over-capitalised.
NPS — surveyed at 30 days post-delivery, tracked rolling 90-day. Target 60+.

Operational health.

Dispatch SLA hit rate (target 95 percent ships within promised window).
Returns rate (target below 4 percent, with QC reprint cost held below 4 percent of revenue).
Agent-fleet approval-queue depth (target below 20 items at any check).
Klaviyo deliverability (target inbox placement above 95 percent, complaint rate below 0.1 percent).
GSC indexed pages and impression trend (target compounding month-over-month from month 3).
Paid ROAS (target 2.5x blended across Meta and Google after month 6, 2.0x acceptable in months 1-6).

Exceptions block. Anything outside threshold is flagged at the top of the digest with the agent that surfaced it, the suggested action, and a yes/no/escalate trigger. Steve clears the exceptions block before 9am AEST every Monday.

Closing Note

The M- thesis is not "can a 3D-print lamp brand exist." Gantri and Wooj have already answered that. The thesis is "can a 3D-print lamp brand be built and operated by an AI agent fleet with one human in the loop, faster and at lower marginal cost than a conventionally-staffed equivalent, with a brand premium that holds." If the answer is yes, the second-order opportunity for Ven Agency is not the lamp business. It is the operating model itself: a stack that can be redeployed across other premium D2C verticals, sold as a service to Ven clients, or replicated under sub-brands at near-zero marginal team cost. M- is the proof.

Final brand name resolves in Phase 10.

The plan, written down.