Private — Do Not Share

Investor Prep Guide

Howard's cheat sheet. Everything you need to know to handle any investor question about Physical Intelligence, PINNs, data, and DePIN economics.

1. Physical Intelligence — The Core Thesis

This is your "why we exist" story. Every investor meeting starts here. You need to explain this in 30 seconds flat.

What It Is

Physical Intelligence = AI that can perceive and reason about the physical world.

Right now, AI (GPT, Claude, etc.) only knows the world through text and images scraped from the internet. It can write essays about heart disease, but it can't tell if YOUR heart is healthy right now. It can describe a city, but it can't sense air quality in real-time.

Analogy: Current AI is like a brilliant person who has read every book but has never left their room. Physical Intelligence gives AI eyes, ears, and touch in the real world — through sensors on real devices carried by real people.

Why It Matters Now

Three trends are converging:

1AI agents need real-world data. Self-driving cars, health monitoring, smart grids — these all need verified physical-world data, not internet scrapes.

2Centralized data is hitting a wall. Apple/Google own the data their devices collect. Users get nothing. Regulations (GDPR, etc.) are tightening. Decentralized ownership is the future.

3Hardware is cheap enough. The sensors in a $160 phone today were $10,000 lab equipment 10 years ago. Consumer devices can now produce meaningful data.

How to Explain It (30-second pitch)

"GPT was trained on the internet — text, images, web pages. But the next wave of AI needs to understand the physical world — health, environment, movement, energy. That requires real-time sensor data from real devices in real places. We build the hardware that collects this data, and the verification layer that makes it trustworthy. 25,000 devices deployed, $4M revenue, zero funding."

Say This

Don't Say This

2. PINNs — Your Technical Moat

PINNs is the hardest topic. You don't need to understand the math. You need to understand the logic and be able to explain it with an analogy.

What PINNs Is (Simple Version)

PINNs = Physics-Informed Neural Networks. A type of AI model that "knows" the laws of physics.

Normal AI/ML: Learns patterns from data. "Heart rates between 60-100 are normal."
PINNs: Learns patterns from data AND enforces physical laws. "This heart rate reading must be consistent with this person's blood oxygen, body temperature, and movement — according to the laws of cardiovascular physics."

Analogy: Normal ML is like a student who memorized the answer key. PINNs is like a student who understands the underlying physics — they can't just memorize wrong answers, because wrong answers violate the physical laws they've internalized.

Why PINNs Matters for Oysterworld

Problem: Consumer sensors are noisy and can be spoofed. If you want to sell health data to pharma companies, they need to trust it's real.

Solution: PINNs cross-validates multiple sensor readings against physical laws. Heart rate x blood oxygen x movement x temperature must all be biologically coherent. If someone tries to fake data, they'd need to simultaneously fake ALL sensors in a way that's physically consistent — which is computationally impossible.

Analogy: It's like a lie detector that checks your story against the laws of physics. You can lie about one thing, but you can't lie about everything consistently without violating a physical law somewhere.

The Key Number: MAE 0.35

MAE (Mean Absolute Error) = 0.35. This means our PINNs model predicts sensor values with an average error of 0.35 units.

Clinical threshold is 0.61. Anything below that is considered clinically acceptable.

Our verification is nearly 2x more accurate than required. This is the number investors will remember.

If asked "where does 0.35 come from?": "We benchmarked our PINNs implementation against the methodology published by Raissi et al. in the Journal of Computational Physics (2019). It's the foundational paper in this field — over 8,000 citations. Our implementation achieves 0.35 MAE on cross-modal biometric verification."

PINNs in 3 Sentences (Memorize This)

"PINNs stands for Physics-Informed Neural Networks. Unlike regular AI that just learns patterns from data, PINNs encodes the actual laws of physics — thermodynamics, biomechanics, fluid dynamics — into the model. This means our data verification can't be fooled, because fake data would have to simultaneously violate multiple physical laws to get past it. Our accuracy is 0.35 MAE against a clinical threshold of 0.61."

Tricky follow-up: "Did you build the PINNs model yourself?" Hard
Honest answer: "PINNs is an established methodology from computational physics. We didn't invent the concept — Raissi et al. published the foundational paper in 2019. What we built is the implementation layer that applies PINNs to consumer sensor data verification at scale. The academic research proves the methodology works. We engineered it into a production system that runs across 25,000 devices."
This is honest and strong. You didn't invent neural networks either — you applied them. Same principle.

3. DePIN Economics — The Business Model

What DePIN Is

DePIN = Decentralized Physical Infrastructure Networks.

Traditional model: Company builds infrastructure, owns it, profits from it. (AT&T builds cell towers.)
DePIN model: Users buy devices, contribute to the network, share in the value. (Oysterworld users buy phones, contribute sensor data, earn rewards.)

Analogy: Airbnb doesn't own hotels — hosts do. Uber doesn't own cars — drivers do. DePIN applies this to data infrastructure. We don't own sensor nodes — users do. But we built the platform that makes their data valuable.

Your Revenue Model (Memorize This)

1Hardware sales (NOW — $4M): User buys device. We earn 60% gross margin. This is the engine that builds the network.

2Data activation (H1 2026): Once devices are deployed, users opt-in to share anonymized data. We take a platform fee on each data transaction.

3Gas fees (H2 2026): Every data verification on the PINNs engine costs a small fee. Like Ethereum gas fees but for data verification.

4Data licensing (2027): Aggregated, PINNs-verified datasets sold to enterprises (pharma, insurance, smart city).

Honest framing: "Revenue today is 100% hardware. That's by design — you can't sell data from a network that doesn't exist. We built the network first. Data monetization is the next phase, but even without it, we have a profitable hardware business."

Why DePIN vs Centralized

Investor question: "Why not just build a centralized data platform?"

Three reasons:

1Distribution: DePIN lets users fund network expansion by buying devices. We don't need to deploy infrastructure ourselves — users do it for us and pay us for the privilege.

2Data quality: Decentralized networks are harder to manipulate because no single entity controls the data. PINNs cross-validates across independent nodes.

3Regulation: GDPR/HIPAA are pushing toward user data ownership. A centralized model is a regulatory liability. DePIN is regulation-forward.

4. Competitive Positioning

The Competitive Landscape (Know These Names)

Helium — DePIN for wireless connectivity (LoRaWAN). Token crashed 94%. Key lesson: they had token before revenue. We have revenue before token.

DIMO — DePIN for car data. $300M+ FDV. Proves that physical data networks can achieve real valuations.

World (Worldcoin) — Biometric identity verification using iris scanning. Raised $1.23B. Zero revenue, zero deployed consumer devices. $5B+ valuation.

Figure AI — Physical AI through robots. $2.6B valuation. Different approach (robots vs sensor network) but same thesis: AI needs physical world data.

Apple Health / Google Fit — Centralized, walled garden. They collect data but users don't own it and can't monetize it.

Your Positioning Statement (Memorize)

"Helium proved DePIN distribution works but failed on economics. World proved biometric data has massive value but is centralized and privacy-invasive. We combine DePIN distribution with PINNs-verified data quality — and we're the only one that's already profitable. $4M revenue, $0 funding, 25K nodes live."

5. The 20 Most Likely Questions

Ranked by probability. Green = easy, Orange = medium, Red = prepare carefully.

1Tell me about Oysterworld in 60 seconds. Easy
"We build the sensory infrastructure for Physical Intelligence. Think of us as the eyes and ears that let AI understand the real world — health data, environmental data, spatial data. We sell hardware devices — phones, wearables, glasses — that collect this data, and we verify it using physics-based AI called PINNs. $4M in revenue, 25K devices deployed, 60% margins, zero funding raised. We're raising our first institutional round — $1M SAFE at $30M post-money — to scale distribution and launch data monetization."
2Why should I care about this market? Easy
"AI companies have spent $100B+ training models on internet data. That data is running out — every major LLM lab is hitting the same ceiling. The next frontier is physical-world data — real-time, real-place, real-sensor data. That's a $50T infrastructure opportunity. We're building the picks and shovels."
3How is this different from IoT? Medium
"IoT is plumbing — it moves data from sensors to servers. We add two layers on top: verification (PINNs ensures data is physically real, not spoofed or noisy) and ownership (users own their data through DePIN, not a corporation). IoT is dumb pipes. We're verified, user-owned intelligence."
4How do you make money? Easy
"Today: hardware sales. $160 average price, $64 COGS, 60% gross margin. 25K devices sold. Next: data activation revenue per active node — launching H1 2026. But here's the key: hardware alone is already a profitable business. Data revenue is upside."
5Who are your customers? Easy
"Two-sided market. Device buyers are crypto-native consumers in SEA and LATAM who want to participate in DePIN networks — they buy the hardware and earn rewards for contributing data. Data buyers (Phase 2) are pharma companies, insurance actuaries, and smart city planners who need verified real-world data."
6Why $30M valuation? Hard
"At $4M revenue, $30M is a 7.7x multiple. For a pure hardware company, that's aggressive — I agree. But we're not pricing on hardware alone. We're pricing the network — 25K deployed nodes already collecting data, with data monetization launching this year. Comparable DePIN networks trade at 20-50x. World Labs raised at $5B with zero revenue and zero devices. At $30M, you're getting hardware profitability as the floor and network effects as the upside."
7Solo founder risk? Medium
"Valid concern. Here's my answer: I have a 32-agent AI development factory that produces code 24/7 across 13 compute nodes. It's not hypothetical — it's how I shipped 5 products, generated $4M revenue, and built a 140K+ community. The factory IS my team. That said, I'm actively hiring key technical talent for the data monetization phase. The $1M raise gives me runway to build out the team."
This is honest and turns a weakness into a story about resourcefulness. Shows self-awareness about scaling needs.
8Explain PINNs to me like I'm 5. Medium
"Imagine you have a kid who says 'I ran a 3-second 100m dash.' Regular AI checks: is 3 seconds a number? Is it in the right format? Yes? OK, accepted. PINNs checks: is it physically possible for a human to run 100m in 3 seconds? No — the fastest human ever did it in 9.58. Rejected. PINNs applies the laws of physics to every piece of data. If it's physically impossible, it gets flagged."
This analogy works for anyone. Practice it until it flows naturally.
9What's your CAC? How do you acquire customers? Easy
"Near-zero — effectively $0. All 25,000 devices sold through community channels: Twitter/X, Telegram groups, DePIN communities. Zero paid advertising. Zero influencer deals. The product sells itself because the economics are transparent: buy a $160 device, earn rewards for contributing data. As we expand into new markets and product lines, we budget for marketing, but to date our acquisition has been 100% organic."
10What's your unfair advantage? Medium
"Three things. First: 25K deployed nodes — that's a cold-start problem solved. Building a DePIN network from zero is the hardest part. We've done it. Second: PINNs verification — no other consumer hardware company has a physics-based data verification layer. Third: AI Agent Factory — I can ship at 13x the speed of a traditional team at near-zero cost. Our cost structure is a permanent advantage."
11Who is actually going to buy this data? Hard
"The first buyer will be pharma companies running Real-World Evidence (RWE) studies. The FDA now accepts RWE for drug approvals — it's a $2B+ market growing 15% annually. They currently pay $500-2000 per patient for clinical trial data. We can offer PINNs-verified biometric data from thousands of users at a fraction of the cost. That said — we haven't closed a data deal yet. That's the next milestone, and it's what this raise funds."
Being honest about "haven't closed yet" builds trust. Showing you know the buyer and market shows preparation.
12What happened with the Stripe reversals? Hard
"Transparent answer: the $1M in Stripe reversals was from a single distributor deal in 2024 where the terms were restructured mid-fulfillment. The net on that deal was $750K. It's not fraud, chargebacks, or customer dissatisfaction — it was a contract restructuring. All other Stripe transactions are clean. I can show you the Stripe dashboard live."
Have Stripe dashboard ready to show. Transparency here is critical — if you dodge this, they'll assume the worst.
13Why crypto payments? Is this a crypto company? Medium
"45% of our revenue is in TON crypto because our largest customer base is in SEA and LATAM where crypto is the preferred payment method. It's not an ideology — it's distribution optimization. We accept fiat (Stripe) and crypto (TON, MixPay) because our customers use both. The product works regardless of payment method."
14What do you use the AI Agent Factory for exactly? Medium
"Everything a traditional engineering team does: writing code, running tests, deploying, fixing bugs. 32 AI agents across 13 compute nodes, running 24/7. It costs me roughly $100K/year in compute — that's what one junior engineer costs. The factory shipped 5 products, this entire investor website, and our data verification pipeline. It's not a demo — it's production infrastructure."
15What's the token plan? Medium
"$OYS token is Phase 2 — after this round, after data monetization is live. We're deliberately not launching it now because we saw what happened to Helium: token before product-market fit = speculation bubble then crash. Our approach: revenue first, token second. When we launch, the token will have real utility — staking for data validation, rewards for node operation, payment for data marketplace transactions."
16How do you ensure data privacy? Medium
"Three layers. User control: users explicitly opt in to share data, and can revoke anytime. Anonymization: data is stripped of personally identifiable information before it enters the marketplace. On-device processing: PINNs verification runs on the device itself — raw biometric data never leaves the phone. Only verified, anonymized data points are shared."
17What are the risks? Easy
"Three main risks, in order: 1) Data monetization execution — hardware is proven, data sales are not yet. That's why I'm raising. 2) Solo founder dependency — I'm mitigating this with the AI factory and actively hiring key technical talent. 3) Crypto market correlation — 46% of revenue comes through crypto channels, so a major crypto downturn could slow sales. I'm diversifying toward fiat channels."
Investors love founders who name their own risks. It shows self-awareness and builds trust.
18What milestones will this $1M unlock? Easy
"Four milestones in 12-18 months: 1) 50K cumulative nodes deployed. 2) First $1M in annual recurring data revenue. 3) First enterprise data deal (pharma or insurance pilot). 4) Key technical hires onboarded. These collectively prove the thesis and set up a Series A at $50M+."
19Why raise now? You're already profitable. Medium
"We're profitable on hardware, but growth is constrained by go-to-market spend. Near-zero CAC means 100% of hardware margin is retained. The $1M isn't for survival — it's for acceleration. Without it, we grow 30% year-over-year on organic. With it, we can 3x by funding marketing, BD, and the data activation launch."
20What happens if data monetization doesn't work? Hard
"We survive. Hardware at 60% margins with near-zero CAC is a standalone business. We'd continue scaling hardware — phones, wearables, glasses — across new markets. The company doesn't die if data monetization is slow. It just grows linearly instead of exponentially. But I believe the data thesis is real, and $1M gives us the runway to prove it."
This is the most important answer. It shows the downside is "a profitable hardware company" not "zero." That's the safety net investors need.

6. Body Language & Delivery Tips

The 3 Rules

1Lead with numbers, not vision. "$4M revenue, zero funding" hits harder than "we're building the future of Physical Intelligence." Open with proof, then paint the vision.

2Say "I don't know" when you don't know. Follow it with "but here's how I'd find out" or "that's exactly what this raise would let us test." Investors respect honesty over BS.

3Pause after big numbers. After saying "$4M in revenue with zero funding" — stop. Let it land. Don't immediately rush to the next point. Silence is powerful.

If You Get a Question You Can't Answer

Template: "That's a great question and I want to give you an honest answer rather than make something up. [What I do know is X]. [What I'd need to research is Y]. Can I follow up with you on that specific point by [date]?"

This works because it shows integrity, demonstrates what you DO know, and creates a follow-up touchpoint (which investors like — it shows professionalism).