P34 Membership · Founding cohort open

Turn your AI agent into a business operator

Your AI can talk. P34 lets it do business. Membership is direct access to the research platform for robots that make money: a persistent workspace, economic decision models, market blueprints, compute — and progressively more autonomous execution, alongside a community exploring what AI can actually operate on our behalf.

Sign up at the console, subscribe, create an API key, connect your agent. The simulator is open the moment you're in.

What a member session looks like
Deal on offerAskP34 expectedCall
1962 Seiko chronograph$410+$120Buy
Omega parts lot (14 pcs)$1,250−$180Skip
Citizen NOS × 4$300+$95Buy
“Rare” Rolex bezel$890−$340Skip
Tag Heuer, 1980s$520+$60Buy
Vintage Casio bundle$240−$25Skip

3 buys. 3 refusals. The refusals are the product — anyone can spot an upside; the money is in not paying for the other three. Illustrative example.

Included today

What members get now

Present utility first. Everything below ships with the membership — starting with a virtual machine that never goes to sleep. No waitlist theater.

24/7 virtual machine

An always-on managed VM for your agent, preloaded with market-access tools, curated data sources and web scraping — so it can collect data and operate the business continuously, between your sessions and while you sleep. HyperC-operated agents keep it running.

Market access

Point P34 at the market you already operate in — that is the recommended path. Maintained workflows exist where we have execution and data coverage (Amazon wholesale, U.S. and Europe), but a maintained workflow is not the same as an easy market, and your own market beats a pre-built one. See the catalogue of 135 experiments and blueprints.

Model & compute credits

Individual calculations are expensive. Membership converts part of the fee into a defined monthly compute allowance — you see usage as a simple percentage.

API & agent skills

The P34 REST API, agent skills, implementation examples and private technical materials — built to be called by Claude, ChatGPT, open models or your own code.

Simulator & sandbox

A browser market simulator at api.hyperc.com/sim plus test keys and sample datasets — build intuition before wiring real data.

Opportunity feed

Periodic model-generated commercial opportunities that meet published P34 acceptance criteria, offered to eligible members under defined allocation rules.

Community & research

The community is where operators, builders and researchers share market experiments, operating knowledge, results, lessons and evidence — and the open questions about how increasingly autonomous businesses should behave.

Bring your own agent

P34 is the substrate. The agent is yours.

Claude, ChatGPT, open-source models or custom code — your agent orchestrates; P34 supplies the one step it cannot do alone: the economic decision.

OwnerObjectives, constraints, capital. You stay in charge of all three.
AgentClaude / ChatGPT / open-source / custom — reads markets, builds menus, talks to you.
P34Models, data, skills, compute — scores the menu, refuses the losers, sizes the rest.
MarketSupported commercial workflows with real execution paths.
ResultOwner-controlled transaction and outcome. Your business, your economics.
Developer primitives
Opportunity searchFind candidate commercial actions in your market
Economic scoringEstimate contribution margin, cash use, risk, uncertainty
Constraint checkingReject candidates that violate owner or model limits
Scenario simulationStress-test price, fees, demand, returns and delays
Action preparationPrepare an order or workflow for your approval
ExecutionExecute only where enabled and authorized by you
The API, in one breath
# POST https://api.hyperc.com/v1/fit
r = requests.post(API + "/fit",
    headers={"Authorization": "Bearer " + KEY},
    json={"menus": menus,      # every option, incl. declined
          "sales": sales,      # realized outcomes log
          "market_type": cfg}) # your market's economics
sid = r.json()["session_id"]
# poll GET /result/{sid}: queued → processing → done
# done ⇒ your T=0 menu, filled in:
#   qty per key (0 = do not trade) + predicted profit

Full docs, examples and the pytest integration workflow →

A member workflow, end to end

Discover → score → reject → select → execute or hand back

01DiscoverYour agent pulls a live menu: wholesale lots, listings, loan applications — whatever your market offers today.
02ScoreOne API call. P34 calibrates to your history and prices every option — including the ones you'd never have looked at.
03RejectMost options fail. In the published benchmark study, P34 rejected 99.4% of evaluated orders. Refusal is the discipline.
04Select & sizeSurvivors come back as a sized portfolio with predicted economics, calibrated as a sum — the book, not just picks.
05Execute or hand backYour agent prepares the orders. You approve — or, where enabled and authorized, it executes within your limits.
Market access

Deployments, experiments and blueprints

Pick your market before you read this list

The market to start on is the one you already operate in, or know well. The state on each entry records where P34 has already been pointed — evidence, not a ranking and not a recommendation. Amazon wholesale is the reference deployment and one of the hardest markets to enter — Amazon account management and wholesale supplier relationships are demanding operating problems that P34 does not solve. Bring your own market and its telemetry: 135 are catalogued, and the fit check tells you in a minute whether yours qualifies.

Reference deployment

Amazon wholesale — US & EU

The founding deployment: purchase and shipping portfolio decisions from live wholesale menus, validated in production at ~$100M/yr scale. Developed, but hard to enter — account and supplier management are yours to run.

Reference deployment

Synthetic markets — simulator

Open to every member: configurable synthetic markets in the browser and via API, for learning and integration testing. Not real money.

Active experiment

Collectibles & online arbitrage

Domains, vinyl, retro games, watches, TCG — dense public telemetry, small tickets. Experimental workflows; bring your own execution.

Active experiment

Micro-lending

Validated in a ~3,000-loan live model test. Gated as regulated commerce: separate legal review before member access.

Market blueprint

Freight, leads, capacity

Load boards, local lead arbitrage, GPU capacity — menu-shaped and computable on paper; telemetry validation in progress.

Market blueprint

Your market?

Run the market-fit check. If it's computable and you can bring the telemetry, we want to hear about it.

Financial-market uses (securities, derivatives, certain crypto activity) sit in a separate product and regulatory perimeter and are not part of the membership offer. Regulated commerce markets are enabled only after market-specific legal review.

Founding economics

Why joining early is contractually better

Founding advantages here are the ones written into the terms — not vibes. The Founding Operator Beta is limited to a maximum of 1,000 paid accounts.

Success-fee lock

Where profit-share pricing applies, the earliest paid accounts receive the introductory 10% performance-fee rate — assigned by paid-registration order and shown in your account. Later cohorts pay more (up to 30% under standard terms).

Founding price

Your $2,000/month rate holds while you remain eligible and active. Members on an earlier plan keep the rate they joined at. Price increases require 30 days' notice and apply only from your next renewal — cancel any time before.

Priority access

First access to new model capabilities, new market experiments and additional capacity as they release. Opportunity-feed eligibility follows published allocation rules.

Straight talk. You own and operate your business and receive its economic result. P34 supplies software, intelligence, compute and execution infrastructure, and charges subscription, usage and — only where you enroll a project — success-based service fees. Membership is not an investment, not equity, and not a promise of income. See the Membership Terms.
Autonomy roadmap

Current vs planned capability

Four rungs of the same research question: how far can the owner move from operating every step to setting objectives, constraints and oversight? The labels say what ships now versus what is planned.

Available now

1 · Assist

Data, APIs, research and model-scored opportunities. You decide and execute.

Available now

2 · Copilot

Your agent assembles analysis, compares actions and prepares a recommended workflow for your approval.

Rolling out

3 · Supervised autonomy

Defined actions prepared or executed inside owner-set constraints, with supervision.

Planned

4 · Self-driving business

You set objectives, capital and risk limits; agents automate increasing portions of operation. Direction, not a current promise.

Evidence & risk

Don't believe us. Compute it.

We publish methodology, benchmark results, losing cases and limitations — and invite you to challenge them with your own AI before you subscribe.

Synthetic stress test
+$2,250
realized on $4,146 deployed — P34, withheld task menu

Executed REST-API run, week-80 menu of a synthetic market. The tuned XGBoost baseline realized −$8,789 on $18,821. Synthetic environment: not evidence of live-market profitability.

Synthetic stress test
−$417.4k
what the "great backtest" baseline lost on the full future menu

A 0.9266 ROC-AUC classifier looked profitable on its observed holdout (+$61.7k) and collapsed out-of-menu. This is the trap P34's rejection discipline exists to avoid.

Live deployment
$30M+
sales generated for customers, with >95% of trades unsupervised

Company-reported, cumulative, Amazon wholesale. Plus ~3,000 loans issued in a live lending test. Note: not audited by a human licensed auditor; historical results don't guarantee yours.

Pricing

One membership

A single plan while the Founding Operator Beta runs — the full compute allowance, no tier to choose. Enterprise deployment is a separate, custom conversation.

Members who joined on an earlier plan keep that plan, its price and its allowance until they cancel — nothing changes on an existing subscription.

Enterprise

Custom limits, custom retention, SLAs, security terms, staged rollout and governance. Pricing by Order Form.

Explore Enterprise →

Membership is subject to fair-use limits (a weekly burst limit on top of the monthly allowance) and the Membership Terms + API Terms of Use. Future capabilities are planned, not guaranteed. Prices exclude taxes.

Founding members get there first.

Seven years of decision-model research, one $2,000 membership. Connect your agent this afternoon.