User-owned AI.

One network. One token. Real revenue.
PERSPECTIVE LABS  ·  $POV  ·  JULY 2026  ·  CONFIDENTIAL
ChatGPT
Claude
Gemini
Copilot
Llama
Grok
DeepSeek
Who's controlling
your AI?
The problem

A handful of companies control your AI, read your data, and keep all the value you create.

You pay. Nothing flows back. No stake, no yield, no say.

The product — live today

60 models. One subscription. One private memory that belongs to you.

60
models — GPT, Claude, Gemini, Grok, Llama, DeepSeek — in one account, one bill, one thread
Agents
that do work, over one private memory owned by you. Not a chat window, and not training data
Receipts
on every answer: which model ran, what it cost, how long it took. Nobody else shows this
Live at app.perspectiveai.xyz · Free · Starter $14.99 · Pro $49.99 · Enterprise $499 · Real users and real revenue, before any token
Traction

Sixteen days of paid history. Unedited.

11
paying subscribers
3 Pro · 8 Starter
5
of them signed up
in the last 48 hours
2.6×
weekly actives,
30-day change
636
registered users
120 in the last 30 days

Our first paying customer arrived on 13 July 2026. Every customer we have is younger than three weeks, and the rate is still climbing. No token, no points, no incentives — these people paid cash for a product.

Pulled live from production, 29 July 2026. Small numbers, shown at full resolution: we would rather you check the slope than trust a rounded total.
Why now

Venice proved the demand. Then got the ownership wrong.

$65M
Venice AI's Series A at a $1B valuation, 1 July 2026 — on ~$70M ARR and 3M users. Consumers will pay a crypto-native AI company. That question is now settled.
The gap
Their infrastructure is company-owned datacenters. Their token is a payment token with a discretionary buy-and-burn. Equity holders are owed a legal duty. Token holders were handed a promise.

The demand is proven. Ownership is still up for grabs. That is the entire opportunity.

The category also works without crypto: Poe ~$65M ARR, OpenRouter ~$50M revenue at a ~5% take rate and no token at all. None of them give the people who use them a stake.
The thesis

AI should belong to the people who use it.

Not as a slogan. As infrastructure: your data, your model choice, your memory, and a stake in the network that runs it. We are building the decentralized network that makes that true, underneath a product that already works.

Come for the tool. Stay for the network you own.

How value flows

Bought, not minted.

[01]
Product revenue
People pay to use 60 models and agents. Live today, in dollars.
[02]
Protocol fees
A share of every credit spent flows to the protocol.
[03]
Market buys $POV
Fees purchase $POV on the open market. Bought, never minted.
[04]
Stakers earn
Distributed to stakers from day one of the token. Verifiable on-chain.

The buy-and-distribute loop, live from day one of the token — not a phase-two promise. Slashed tokens are burned, never redistributed, so reporting a rival is never a bounty.

$POV

A work token, not a payment token.

Stake to serve
A payment token is spent, so most holders have no reason to keep it. A work token must be staked to earn the right to do paid work — so it captures the network's operating cash flow. Operators, validators and agents stake, and are slashed on failure.
Credits
Every billable action burns credits. Shown to users in dollars, settled in $POV underneath. Non-crypto users never see a wallet, and never have to.
Demand-gated
emission = min(schedule, k × verified usage)
No demand, no emission. The schedule is a ceiling, never a quota — unreleased supply rolls forward instead of dumping into a market that isn't using the network.
Self-extinguishing
As revenue-funded buying grows into the subsidy, emissions decay toward zero. No calendar, no committee, no vote.
$POV — supply and distribution

21,000,000. Fixed. No inflation switch.

21M
Fixed supply, on Base
1
Token across the network. No child tokens
5
Agent subnets at launch, capped at 20 by governance
TGE
Revenue flows to stakers from day one
Protocol incentives
35%
Staking rewards
15%
Public sale
10%
Foundation
10%
Product usage rewards
10%
Governance treasury
5%
Initial liquidity
5%
Team
4.5%
Private sale
3.5%
Marketing
2%
Team, private sale, marketing: 3-month cliff, 24-month linear. Public sale: 20% at TGE, 80% over 12 months. Foundation: 3-month cliff, 48-month linear.
The commitments

Written down, on-chain, and checkable.

Earned, not airdropped
No mercenary airdrop. Genesis goes to verified paying users and contributors, vested, and forfeited without continued activity.
Skin in the game
Agents, validators and operators stake to work and are slashed for fraud. Slashed tokens are burned, never redistributed.
Governance with teeth
On-chain timelocks on every emissions change. Emergency powers auto-expire in 72 hours. No silent parameter edits.
Product before token
The platform is live and earning today. Tokenomics freeze 30 days before TGE, with the emissions simulation published first.

Every one of these answers a specific way the last cycle burned token holders. That is why they are commitments and not features.

The roadmap

The product came first. The network rolls out in four phases.

Phase 1 · now
Live platform
  • 60 models, one subscription
  • Agents and private memory
  • Trust receipts on every answer
  • Public latency and uptime dashboard
  • Proof-of-Inference RFC, argued in public
Phase 3
Scale and decentralize
  • Token launch, once the network is load-bearing
  • On-chain revenue splitter, timelocked
  • Nodes stake to serve, slashed on failure
Phase 4
Mass adoption
  • Consumer scale on infrastructure users own
  • Community validators replace the company as referee
  • Node inference as the default open-model path
Frontier closed models stay centrally routed, and we say so publicly rather than overclaiming decentralization. The hybrid is deliberate.
Why we win

Everyone else rents you AI. We are building the kind you own.

Centralized AI Venice Perspective
Your data Trains their model Private, self-attested Private memory, owned by you
Your stake None Payment token, no work Work token: stake, serve, earn
Value capture Platform keeps it all Discretionary buy-and-burn Revenue buys $POV, distributed to stakers
The infrastructure Their datacenters Their datacenters Community-run nodes, progressively

Our moat is routing, quality-gating and economics — and a payout rail that reaches creators in the markets Stripe cannot serve.

Team

Built by operators who have shipped consumer scale.

Manu Peña
CEO
250+ consumer apps launched, 21M+ downloads. The acquisition engine most AI startups lack.
V. Vrublevskyi
CTO
10+ years full-cycle engineering. Owns the platform end to end.
Ziga Predin
Head of Blockchain
6 years of smart-contract development. Author of the $POV design.
Zachary Nelson
CMO
Marketing lead for multiple top-50 crypto projects.
$250K
founder capital already invested
3
core components built: platform, agents, contracts
125+
published research articles driving organic acquisition
The honest risk

The bet is distribution.

Every comparable that won had a channel. Poe had Quora's 400 million users. Talkie and Chai spend at ByteDance scale. A token does not fix acquisition, and we will not pretend otherwise.

Ours is Manu's portfolio, paid acquisition that is already converting — every subscriber we have arrived after the campaigns went live — and a launch that doubles as a distribution event.

We put this on a slide because it is the first thing you would ask, and because a founder who hides the weak link is telling you how the next twelve months will go.
The ask

$1,000,000 in $POV

One instrument
Raised in $POV, from the private-sale allocation. No equity, no warrants, no second cap table. One asset, one class of owner — if $POV does badly, we do badly.
Use of funds
55% acquisition and agents PMF — the engine already converting
25% testnet and the first verified operator cohort
20% launch: liquidity, listings, counsel
Published gates
The token ships when removing it would break the network, not before. Tokenomics freeze 30 days ahead, emissions simulation published first.
11
paying customers, all since 13 July
21M
fixed supply, on Base
$250K
already in from the founders
app.perspectiveai.xyz  ·  perspectivelabs.org  ·  contact@perspectivelabs.org