Consumer first.
Enterprise follows.

The product is built. The infrastructure is live. The patents are filed. We are raising $1M to go to market.

ChatGPT did not start with enterprise procurement. It started with a free chatbot that hit 100 million users in two months. Microsoft invested $10 billion that same month — priced on the growth curve, not on enterprise contracts. Perplexity started as a consumer search tool, now valued at $22.6 billion. Cursor started as a code editor individual developers paid $20 a month for, now valued at $60 billion. Consumer adoption created the valuation. Enterprise revenue came after.

Crypto followed the same pattern. Bitcoin was retail for eight years before the ETF. Ethereum grew because developers built apps and users bought Ether to use them. By the time institutions showed up, the price had already moved by orders of magnitude. If that story is yours, the full parallel to Safebox, Safebots and $SAFEBUX is further down this page.

None of them won by having a better product and waiting. OpenAI burned through billions in compute subsidies to make ChatGPT free. Perplexity raised $300M before it had meaningful revenue. Cursor raised $900M in under four years. The product got them noticed. The capital got them to scale. A better mousetrap sitting in a garage does not become a movement.

That is why we are raising. The product is built. The $1M buys three things:

The $1M also funds listing $SAFEBUX on exchanges and marketing the token to the same audience we are already reaching. A million community leaders around the world, and the people who follow them — we help them run their communities, and they buy $SAFEBUX on exchanges or directly through us to pay for compute, storage and bot capacity. That purchasing creates real demand. Investors who staked their $SAFE tokens earn yields from that demand. Meanwhile the equity valuation grows through the adoption metrics above. Two engines running at the same time: token cashflow for stakers, equity appreciation for everyone.

The distribution assets are real — a million emails, 38K push-reachable devices, five named projects already live. What the capital buys is the marketing, the operator training, the content engine, and the reactivation campaigns that turn those assets into the growth curve the next investor prices on.

What the numbers look like at each checkpoint.

These are targets, not commitments. They assume the $1M pre-seed closes in September 2026 and gives us an eighteen-month runway. The first three months are about getting real — product live, first operators trained, first leaders reactivated, first projects generating content. The compounding starts at month four and accelerates through the Series A at month eighteen.

Step by step, starting with what we already have.

Six million people downloaded the Groups app portfolio. About a million left an email address. 38,000 devices are active in the last twelve months — those users still have the app installed and accept push notifications. Every time we push an update, hundreds of thousands of users receive it. We start here.

Appfigures dashboard — 6M downloads, $463K revenue, 100+ countries. The U.S. alone: $297K revenue, 14M app updates. This is the audience we are reactivating.

Two channels for video. The first is a managed social-growth partner like Viral Coach and the Shane Hummus model — high-volume, searchable content that solves expensive problems for a narrow audience, with a direct call to action into a demo or signup. One video becomes transcripts, clips and posts. The AI safety news cycle gives us fresh material every few weeks.

The second is paid crypto influencer distribution through agencies like AAudience, which packages YouTube channels by region, language, format and audience size. Example inventory from their October 2025 sheet:

We do not need to buy all of these. The International Starter at $4,900 gets ten channels in five languages with a one-week turnaround. Run it, measure signups and activation, and scale what converts. The entire inventory across all packages totals about $170K — a fraction of the $1M raise — and covers English, Russian, French, German, Spanish, Portuguese, Arabic, Hindi, Filipino, Vietnamese, Indonesian, Italian and Greek.

This is the move that makes the rest compound. We do not pay the big influencers for a sponsorship. We build them their own self-sovereign website — a membership site where they charge for access, publish long-form content, host their community, and put teasers on YouTube. Powered by Safebots. Hosted in a Safebox. Their brand, their domain, their audience, their revenue.

They already have the audience. They tell that audience about their new site. The audience joins. We earn on the compute. The influencer is not a channel we are renting. They are a customer whose distribution is their own following.

Each of these people tells their audience: "I have a new site. It is powered by Safebots, hosted in a Safebox. You can have one too. And if you want to run the infrastructure, take the course."

People who see these sites — the influencer's audience, the YouTube viewers, the newsletter readers — some of them want to run the same thing for their own community. Some of them want to go further and run the infrastructure itself. That is the Safebox Operator Course. Graduates incorporate, deploy Safeboxes, onboard customers, and earn on usage and network capacity.

The funnel: influencer promotes their app → audience sees it → some become members (we earn on compute) → some start their own community site (we earn more) → some take the course and become operators (they earn, and we earn on the $SAFEBUX their customers use). Each layer of the funnel feeds the next.

A community leader gets their branded app and bots. Their members join, start interacting with the bots, build history, form connections. That interaction is the lock-in — the members' data, relationships and workflow memory make the system harder to leave every week. Once the community is active, we start charging the leader a recurring weekly fee for compute, storage and bot capacity.

VCs at the next round want to see recurring revenue and hockey-stick growth. This is how we produce both: activate community leaders, let their members create the usage, then charge weekly. Each leader is a small recurring-revenue account, and there are a million of them in the email list.

Try it free. Onboard your community. Then you need it.

We do not have a sales funnel today. That is what the $1M builds. The funnel works like every freemium SaaS that locks in through usage rather than contracts:

Where every dollar goes.

The $1M is an eighteen-month runway. Here is what it buys, in order of priority.

Over half the budget goes to distribution — content production, paid influencer reach, email reactivation, influencer setup, exchange listing and the sales funnel. That is the plan: the product is built, the $1M goes to putting it in front of people.

From $10M to $200M happens fast when it happens.

The January 2027 round opens at $10M. A lead VC investor sets the terms, and the round fills behind them. What makes it oversubscribed is the same thing that made every comparable in the pre-seed document oversubscribed: a growth curve that looks like it is still accelerating when the deck lands on the partner's desk.

Two things VCs need to see. Nothing else matters at the seed stage:

That is all. Not enterprise contracts. Not a finished product roadmap. Not a complete team. A growth curve and recurring revenue. Everything in this document — the YouTube content engine, the influencer apps, the email reactivation, the operator course, the sales funnel — exists to produce those two things by January.

Cognition went from $350M to $2B in one month. Cursor went from $400M to $60B in four years. Perplexity went from $520M to $22.6B in under two years. Once a VC round closes with momentum, the next round reprices aggressively — because the lead investor's reputation is now attached to the company, their portfolio companies become distribution partners, and the press coverage from the round itself drives more adoption.

A $10M seed round that closes oversubscribed in Q1 2027, with a growth curve still accelerating, reprices to $50M at the Series A in early 2028. If the Series A closes with institutional demand and press coverage, the next step to $200M is the same mechanism that took every company in our comparable set from their seed to their current valuation. The math is not speculative. It is the observed pattern across 74 companies founded in the same window.

A bridge to the VC round — and what VCs will see when they look.

The $1M is an eighteen-month runway. It covers the team, the infrastructure, the operator course, the marketing and reactivation campaigns, the patent filings, and the first wave of content and press. The next round — $2M at $10M, opening January 2027 — is where VCs, accelerators and AI-focused seed funds come in. The $1M bridges to that round and funds the adoption metrics they price on.

VCs at the January round will want to see a growth curve. They will want recurring revenue. They will want hockey-stick numbers on a slide. The question for this investor is: where does that growth curve come from in four months?

It comes from adoption. The same place it came from for every company that followed this pattern.

In 2015, Ethereum had no enterprise customers. It had no recurring revenue in the way a SaaS company does. What it had was a developer community writing smart contracts, a user base buying Ether to pay for gas, node operators running validators, and a price chart going up because all of those things were growing at the same time. When institutional investors looked at Ethereum in 2016 and 2017, they did not see a revenue waterfall. They saw a network growing faster than anything they had seen before, and they priced accordingly.

The same thing happened with OpenAI. ChatGPT launched as a free consumer product in November 2022. By January 2023 it had 100 million users. Microsoft invested $10 billion that same month. The $10 billion was not priced on enterprise contracts. It was priced on the growth curve — a hundred million users in two months, the fastest consumer adoption in history.

Perplexity, Cursor, Midjourney — same pattern. The VC round priced on adoption velocity, not on the P&L.

The pre-seed investor's position goes from 25% at $4M to roughly 20% of $10M after the January round — about $2M on paper, a 2× in four months. That is not a liquidity event. The first realistic window to convert any of it to cash is the Series A secondary. But the valuation step from $4M to $10M is what the $1M is designed to produce, and it is driven by the same thing that drove every comparable in the set: adoption growing faster than a VC's spreadsheet predicted.

Read it plainly — the $1M buys four months of aggressive consumer activation. The VC round prices on what that activation produced. Enterprise revenue is the long game, and it is real, but it is not what gets us from $4M to $10M. Adoption is.

One architecture, three things an investor should care about.

Most AI companies sell one thing. Inference brokers sell cost savings. Guardrail vendors sell safety. Agent platforms sell ease of use. Each trades against the other two — cheaper means fewer checks, safer means more overhead, simpler means hiding the controls.

Safebox does not have that tradeoff. All three come from the same architectural decision: work is a declared workflow on a sealed substrate, and the model gets called only where judgment is needed.

The thesis underneath all of it: capability is model × substrate. The entire AI industry is spending billions pushing a 9 to a 9.5. The substrate — the thing the model multiplies against — is sitting at a 2, and almost nobody is working on it. Double the substrate and you double the product. That is what we are building, and the cost savings, the safety, and the simplicity all fall out of the same decision.

Every dollar the AI industry spends goes to making the model smarter. Think of model quality on a scale of one to ten. The frontier labs have pushed it from a seven to a nine, spending tens of billions to do it. Getting from nine to 9.5 will cost even more, for half the improvement. Diminishing returns.

Now think about the other factor — the system around the model. The harness, the tools, the workflows, the data it has access to, the domain knowledge. Call that the substrate. On the same one-to-ten scale, the substrate across the industry is sitting at about a two. Nobody is seriously working on it because everyone is fixated on the model.

Capability is not model plus substrate. It is model times substrate. Nine times two is eighteen. Pushing the model from nine to 9.5 gets you from eighteen to nineteen — a 6% improvement, costing billions. Pushing the substrate from two to four gets you from eighteen to thirty-six — a 100% improvement, at a fraction of the cost. We are building the substrate.

Every other claim on this page falls out of that one decision. If most of the work runs on the substrate instead of the model, you stop paying model prices for it — that is the 20× cost savings. If the substrate decides the steps and the model only fills in judgment, the sequence is declared in advance and gated — that is the safety. If the substrate is a workflow rather than a harness you assemble from parts, there is nothing to configure — that is the ease of use. One architectural choice, three commercial consequences. The full argument is in the substrate thesis.

For investors who know blockchain.

If you lived through Bitcoin, Ethereum, Solana and Polygon, the pieces of this company map directly to what you already understand. Safebox is the blockchain. Safebots are the applications. $SAFEBUX is the gas token. Operators run sealed boxes the way validators run nodes. The table and the five-step growth story below lay out the mapping in detail. If crypto is not your world, skip to the ask — everything above already made the case.

If you understand why Ether went from $0.30 to $4,000, you understand the $SAFEBUX thesis. The token is worth something because people need it to use the network. More users, more demand for the token. The network grows, the token appreciates. The difference is that instead of a ledger of financial transactions, this network runs AI — private, verifiable, 20× cheaper AI.

The playbook that took blockchain from zero to a trillion.

Crypto followed a specific sequence that repeated across Bitcoin, Ethereum, Solana and every successful L1. If you lived through it, this will feel familiar. If you are evaluating Safebots, notice that we are following the same sequence.

$1M at $4M. Two ways in.

U.S. accredited investors come in as equity under Regulation D. Eligible non-U.S. investors take the same economics offshore under Regulation S, in a form that can later be tokenized as $SAFE and traded on a FINRA-registered ATS. Both fund the same balance sheet. They differ in who can buy and what liquidity looks like afterward.

The financing ladder: $1M at $4M now (25%). $2M at $10M opening January 2027 (20%). $15M at $50M opening February 2028 (30%). At the Series A we target a company-approved secondary where early investors can recover principal and keep most of the position riding.

The full pre-seed terms, the 74-company comparable set, the measured cost data, and the liquidity path are in the pre-seed document. The investor data room is at safebots.ai/overview.

Read it the other way — if you understood why running an Ethereum node in 2016 was a good idea, you understand why training Safebox operators in 2026 is a good idea. If you understood why Ether appreciated as the network grew, you understand why $SAFEBUX will. The technology is different. The economics are the same.