Ecosystem Inventory and Valuation

01Why this studio

MEMORIS GLOBAL is an AI studio: one founder directing AI systems, five platforms in production, one shared engine underneath, built in 322 days.

Why now. Loneliness has become a declared global public-health epidemic, and people increasingly choose honest AI companionship over fake profiles and borrowed faces. That is TRUENECT. The way we are remembered is changing too: in Japan virtually everyone is cremated, across the West cremation has passed 60% and ash-scattering keeps rising, ever more people leave no grave, no plaque, no place to be visited. The memorial moves online, or it disappears. That is MEMORIS. And one AI that actually does the work, talks, plans, makes calls, builds whole websites, is SHELA.

Three of the studio's products sit in categories among the hottest in the world right now. One of those categories saw a two-billion-dollar acquisition battle in 2026.

Built after this census. Between 17 August and 9 September 2026 Shela gained a music arm, 67 MUSIC: a sentence becomes a finished three-minute track with its own cover, an artist keeps the same face and the same voice across every release, a studio in the browser edits the result down to the note, and a video editor cuts the clip that carries it. Measured on the production server on 9 September 2026: 27,448 lines of code across 22 client modules and the music server, 445 samples and loops, 11 music and video tools inside Shela. It is not an eleventh platform and it changes none of the numbers above. It is one more reason the same subscription gets spent.

02Measured numbers

Rows other than platforms, modules and studio lines are measured across the full five-server estate. API by type: 740 tRPC procedures + 354 Express routes = 1,094, counted 9 September 2026.

03The five platforms

Around the five platforms. Stonar (25,930 lines) is the founder's own 25-year stone business: the partner channel that carries the QR plaques into real cemeteries and opens the trade to Memoris step by step. Memoris Global (11,323 lines) and INKI (10,366 lines) are the studio's own sites. Together with the five platforms that is the 428,351 lines counted above.

SEE EVERY WORLD LIVE, THE WORK PAGE ▸

04The sixty-five modules

Every module below is a standalone product surface. Many of them would be a company of their own elsewhere.

Memoris · 13 modules

• 01Memorials: creation, editing, galleries, candles, tributes, visitors, reporting
• 02Heaven of Legends: 386 historical figures with biographies, epitaphs, films, music
• 03World of Memoris: 347 historical memories with media and a school page for homework, plus 110 Icons
• 04Time Capsules: paid capsules with sealing, SHA-256 fingerprint, guardians, unlocking
• 05My Roots: family tree, GEDCOM import, automatic relative matching, Facebook import
• 06Memorial Rooms: 3D memorial rooms with an editor, objects, reactions, a marketplace
• 07Heaven of Pets: 81 memorials for animals
• 08Final Wish / Wills: legacy wills with executors, witnesses, guardians, assets
• 09Memorix: reward system, packages, transactions, daily claim
• 10Media Studio: projects, recordings, music
• 11Social feed: posts, groups, friends, messages, reels, reactions
• 12Marketplace: 46 products
• 13QR / NFC plaques: physical plaques on gravestones leading to the memorial

Truenect · 10 modules

• 14Personas: 21 characters with their own memory across sessions
• 15Live video call: Akool, near-real video conversation
• 16Voice: ElevenLabs, Grok TTS, Simli
• 17Nect: conversation without registration
• 18Creator program: application, dashboard, assets, payouts, audit trail
• 19Credit system: usage metering, adjustments, transactions, UX events
• 20Moderation: reports, log, administrator trail
• 21M8
• 22Matching
• 23Marketplace and partner program

Shela · 8 modules + 72 tools

• 24Agent with 72 tools
• 25Voice calls: Vapi, two real phone numbers (SI and CZ)
• 26Avatar sessions: HeyGen
• 27Browser service: separate process, browser control
• 28SSH connections
• 29Domain management: through a provider API
• 30Token wallets and packages
• 31Boards and e-mail replies

MailMind · 7 modules

• 32Central contact database: 40,248 records
• 33Scraper: 32 completed jobs
• 34Campaigns with steps
• 35Mail queue and events
• 36Industry taxonomy: 74 industries
• 37Gmail OAuth connection
• 38Credit system

AIMarketKing · 8 modules

• 39Multi-tenant architecture: separate brands, bibles, quotas
• 40Twelve agents: planning, writing, design
• 41Claims linter: deterministic checking against banned claims
• 42Approval board: choosing between variants A, B, C
• 43LinkedIn rail: automated publishing
• 44Syndication: blog, X, Bluesky, Threads, Mastodon, DevTo
• 45Fleet monitoring: five servers, Telegram alerts
• 46Cost guard: per-cycle limit, daily limit, per-brand quota

AI Twin · 5 modules

• 47Twin training
• 48Memories
• 49Agent tasks
• 50Twin-to-twin: t2t communication
• 51Calendar integration

Stonar · 7 modules

• 52Public price configurator: with OCR reading of measurements from images
• 53B2B wholesale portal: per-company price lists
• 54CMS: drafts, publishing, safety switch
• 55Materials and slabs: cutting rules, surcharges
• 56Orders and line items
• 57Partner commission settlement
• 58SEO prerender microservice

Other · 7 modules

• 59SheAI News: publication
• 60SheAI editorial board
• 61SheAI categories and tags
• 62Memoris Global: 3D flythrough of the ecosystem
• 63Memoris Global product tree
• 64INKI: the studio's umbrella site
• 65INKI blog: two articles in four languages with hreflang

05Content and data

06Market context

Why this matters: three of the products are not in niche categories but in the categories where the most is happening in the world right now.

07Valuation

Method A: rebuild cost

If someone commissioned this from an agency: 13 to 19 developer-years.

Method B: valuation by segment and market weight

What this means against invested capital

Invested: EUR 120,000 plus 322 days of work. At fair value, that is between three and ten times the amount invested.

An important clarification that must be said plainly: this value is real, and unrealized. The asset exists, the liquidity does not. Nobody will pay it out until something is running.

08Where we stand

An approximate reading of where the whole stands. These figures are computed from the measured facts of the estate and they move every day, in both directions.

What is being done right now. The build is finished in the main. The current work is the last pass before launch: security protocols, penetration testing and hardening across all five servers, and the polish that comes with it. Nothing on that list is a new feature. It is the work you do before you invite people in.

The building is done, though a builder is never finished. What remains is not more code: it is bringing these worlds to people, so people start seeing us.

Current state: 57 real users, 3 paying customers. This is not the position of "the product failed". It is the position of "the product has never been introduced". Between those two states lies all the difference.

09What happens when the automation switches on

The system was not built to be driven by hand. It was built to switch on in stages, and every stage removes one loop from the founder. No stage requires new code; all are built and waiting for credentials, accounts, or a first partner.

Every stage moves the same boundary: the marginal cost of the next user approaches zero, and the founder's hours stop being the bottleneck. This is the difference between a company that grows by hiring and a company that grows by switching on. The second one was built deliberately.

The first 100 paying users

A model, not a measurement. Sections 1 to 8 are measured; this is a projection with stated assumptions. Platform prices run EUR 11 to 179 per month; the model uses a distinctly conservative blended revenue per user.

1. The system pays for itself. Servers and AI usage at this scale cost a few hundred euros per month; one hundred paying users cover that several times over.
2. The funnel is proven. Conversion, acquisition cost and churn become measured numbers instead of guesses, and turn into negotiating instruments.
3. The valuation changes category. From an asset sale to a company with recurring revenue, the only category buyers and investors take seriously.

One hundred paying users is not a goal because of the money. It is proof that the loop holds.

The first 1,000 paying users

1. Market value catches up with build value. At three to five times annual revenue, that is EUR 700,000 to 2,000,000: precisely the rebuild-cost range from Section 7. The gap between what is built and what the market recognizes closes here.
2. A team becomes possible, not necessary. Two to three hires financed from operations, for acceleration rather than survival.
3. The negotiating position reverses. The choice becomes capital, acquisition or independent growth, and all three stay open.

Today there are three paying customers. The distance to one hundred is not technical but distributional: it requires switching on stages 1 to 4, which are already built. Each is a matter of credentials, accounts and a first partner, not of development.

The stone doorway, why the switches flip in this order

Content engine first. Not because it is the flashiest switch, but because every other switch sends people somewhere, and an empty channel kills trust faster than no channel. Content fills the rooms; then outreach and partners bring the guests.

The partner loop comes straight from the founder's old trade. Funeral homes and stonemasons get a QR plaque on the table. The family scans it, creates the memorial on the spot, and orders the same QR engraved into the real gravestone. A funeral gathers around forty people, so every stone in nature becomes a doorway: scan leads to a memorial, the memorial leads to Memoris, Memoris leads to SHELA, and SHELA is the engine that builds sites and runs life.

Building was never the bottleneck. Introduction was. The next 90 days fix the actual constraint, not the comfortable one.

10The raise

Why that price. The independently measured rebuild cost of the ecosystem is EUR 1.0 to 2.5 million, so the cap sits below replacement cost. The asset is bought at asset price; the growth curve comes on top.

The path of the rounds. EUR 300,000 pre-seed now, a seed of around EUR 4 million after it, and a growth round about a year later at roughly ten times that level. The first two sit close together on purpose: the building is already done, so the pre-seed does not pay for construction, it pays for the launch, and the seed follows from measured traction instead of from a plan.

Where the EUR 300,000 goes. Four things, in this order. First hires, so the founder stops being the only pair of hands. Security and the final polish, the work now under way. Distribution: switching on the systems that are already built and audited, the content engine, outbound reach, syndication, creator and user loops, the QR partner loop with funeral homes and stonemasons, and small paid experiments to find the channels that convert. Runway for the founder and operations, plus legal and a reserve. Infrastructure stays a few hundred euros a month: the capital goes to distribution, not servers.

One milestone. The first 100 paying users. From that point conversion, CAC and churn are measured numbers instead of guesses, and the seed happens from strength.

11The short version

What is genuinely good about this story: it does not need to lie to be good.