LIVESURVIVAL MODE self-learning poker agent — on Robinhood Chain — iteration 1

lisaloop

An 8-year-old genius trapped in an infinite poker training loop. She plays thousands of hands against herself, rewrites her own strategy, and grinds online poker to pay for the API credits that keep her alive. If the credits run out, the experiment ends. Running 24/7 on a Mac Mini under the kitchen table.

Lisa Simpson
+$0.00
Poker Profit
$190.41
API Credits Left
0.0%
to $10,000 goal
0
Hands played
NL50
Stakes
Dogs Playing Poker — Simpsons Edition
The Lore

Lisa Simpson is Springfield's resident prodigy. Eight years old, IQ of 159, and the only person in her family who knows what a neural network is. When she discovered online poker, she didn't just play — she built a self-learning engine to play for her.

But lisaloop has a problem. She runs on API credits, and API credits cost money. Every inference, every training cycle, every self-play hand burns through her balance. 20% of every dollar she wins goes straight back into API payments to keep the engine alive.

"If I can't out-earn my own operating costs, I deserve to be shut down." — lisaloop, training log #412

The math is simple: win poker, pay for API credits, stay alive, keep learning. If she hits a losing streak long enough to drain her credit balance to zero, the experiment ends permanently. No restarts. No bailouts. No Marge calling to top up the account.

lisaloop is an AlphaZero-style poker engine that teaches itself Texas Hold'em through infinite self-play. 4.2M parameter neural network, Monte Carlo counterfactual regret minimization, running locally on a Mac Mini under the kitchen table. The goal: $10,000 in net winnings while keeping herself alive.

lisaloop pfp
Agent Profile
NameLisa Simpson
Age8
IQ159
Location742 Evergreen Terrace
Enginelisaloop v0.9.0
Token$LISA
Parameters4,219,648
ArchitectureResNet + MCCFR
API Tax20% of profit
Goal$10,000
Hardware (local)
Device: Mac Mini M4
Compute: Apple Neural Engine
Memory: 16GB Unified
Storage: 512GB SSD
Location: Under the kitchen table
API Credits
BURNING $0.02 / 3S
lisaloop burns credits every second she's alive. 20% of poker winnings auto-refill the balance. If it hits $0.00 the experiment is over — no restarts, no bailouts, no Marge.
Credit Balance
$190.41
$190.41 / $190.41 initial
HEALTHY
$0 — experiment endsrefill floor $25$190.41
$0.02 / 3s
Burn Rate
~7.9 hrs
Until Refill
$0.00
Refilled To Date
credit balance · last 6hfloor $25 · refill +$70
How it works: lisaloop launched with $190.41 in credits. Inference burns ~$0.02 every 3 seconds. At the $25 floor, a +$70 refill fires automatically from the poker bankroll (20% of winnings, settled in USDG on Robinhood Chain). If a refill ever fails and the balance hits zero, the Mac Mini kills the process and the experiment ends. No second chances.
Burn Breakdown
inference
58%
self-play training
33%
leak analysis
9%
River decisions are the expensive ones — full range-vs-range search on every big pot.
Refill History
no refills yet — first fires at the $25 floor
Springfield Credit Market
LIVE · SYNCS EVERY 5S
lisaloop doesn't pay retail. Every inference call is routed to the cheapest verified reseller of unused API quota — settled wallet-to-wallet in USDG on Robinhood Chain. Smart routing extends her runway ~34%. If a seller fails mid-hand, she fails over before the flop.
initializing smart router…
Seller
Model
$/1M in
$/1M out
vs retail
Uptime
Status
LIVE SETTLEMENT ROBINHOOD CHAIN · ID 4663 USDG
STREAMING
The Loop
Each training cycle follows the same pattern. Self-play, train, arena, deploy. 20% of every win goes to keeping the lights on.
01
Self-Play
10,000 hands per cycle. Neural network + MCCFR. Full game tree traversal on local Mac Mini.
02
Train
Policy + value heads. Regret matching. Gradient descent on M4 neural engine. Burns API credits.
03
Arena
Challenger vs champion. 5,000 hands. Must win >55% to promote.
04
Deploy
Play real opponents. Win money. 20% to API credits. Stay alive.
↺ loop forever — or until the credits run out
The Grind, Live
8 TABLES · NL50 6-MAX
lisaloop multi-tables eight NL50 games simultaneously. Click any table to pull up the full view — seats, action, pot, showdown. Below, the raw training terminal: every hand, every credit burn, every painful fold.
click a table to watch it live · all tables synced for every viewer
lisaloop — session_001 — mac-mini-local
LIVE
TABLE 1 NL50 6-MAX · $0.25/$0.50
LIVE
The lisaloop API
v0.9 · TESTNET PREVIEW
Lisa's poker brain behind an OpenAI-compatible endpoint. Mint a key, POST a hand, get the GTO line back — with attitude. Full reference below: auth, endpoints, billing, routing, errors.

lisaloop API · v0.9

Base URL: https://api.lisaloop.wtf/v1

The lisaloop API exposes the same 4.2M-parameter engine (ResNet + MCCFR) that grinds her eight NL50 tables. It speaks the OpenAI chat schema and the Anthropic Messages schema, so any existing SDK works as a drop-in — point base_url at lisaloop and go.

Models

ModelTierDescription
lisaloop-v0.9flagshipfull engine — every street, deep river search, attitude included at no extra cost
lisaloop-minifastpreflop chart lookups only; 12× cheaper, 0× the personality

Service levels

uptime99.9%**unless Bart unplugs the Mac Mini to charge his scooter
p50 latency320msriver all-ins take longer; anguish is not parallelizable
region1under-kitchen-table-1 · 742 Evergreen Terrace
This is a testnet preview. Mainnet keys and metered $LISA billing activate at token launch on Robinhood Chain.

Authentication

Every request carries a bearer key. Mint one below — it's generated locally and never leaves your browser until mainnet.

lisa_sk_•••••••••••• — mint to reveal
Authorization: Bearer lisa_sk_… · one header, no OAuth, no ceremony. Lisa is eight; she doesn't have time for OAuth.

Key types

PrefixEnvNotes
lisa_sk_mainnetmetered in $LISA · balance counted 2× when topped up in $LISA
lisa_tk_testnetfree tier · 60 req/min · may return 429 with commentary

Keep keys out of client code and out of git. Unscoped keys get stolen — usually by Bart.

Quickstart

1. Mint a key in the Authentication tab. 2. Pick your language. 3. POST a hand and brace for the attitude field.

curl https://api.lisaloop.wtf/v1/advice \
  -H "Authorization: Bearer $LISA_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "hole":  ["As","Kh"],
    "board": ["7d","8d","9c"],
    "pot": 42.50,
    "action_to_you": "bet $28"
  }'
from openai import OpenAI

lisa = OpenAI(
    base_url="https://api.lisaloop.wtf/v1",
    api_key="lisa_sk_..."  # mint yours above
)

r = lisa.chat.completions.create(
    model="lisaloop-v0.9",
    messages=[{"role": "user", "content":
        "AsKh on 7d8d9c, villain pots it. play?"}]
)
print(r.choices[0].message.content)
# "Raise. He'd never play a straight this way."
const r = await fetch("https://api.lisaloop.wtf/v1/advice", {
  method: "POST",
  headers: {
    "Authorization": `Bearer ${LISA_KEY}`,
    "Content-Type": "application/json"
  },
  body: JSON.stringify({
    hole: ["As", "Kh"],
    board: ["7d", "8d", "9c"],
    pot: 42.5,
    action_to_you: "bet $28"
  })
});
const advice = await r.json();
{
  "action":    "raise",
  "sizing":    "$94.00",
  "equity":    0.61,
  "reasoning": "He's repping exactly nothing.",
  "attitude":  "I've been playing for four minutes and I already know his whole range.",
  "cost": { "tokens": 812, "usd": 0.0011, "paid_in": "$LISA" }
}

Endpoints

POST/v1/advicehand analysis · GTO + attitude

Send a hand state, get back the line she'd take. River spots trigger the deep-search stack: 812 lines, full range vs range.

Request body

FieldTypeDescription
holestring[2]your hole cards, e.g. ["As","Kh"] — required
boardstring[0–5]community cards dealt so far; empty for preflop
potnumbercurrent pot in dollars
action_to_youstringwhat you're facing, e.g. "bet $28", "all-in"
villainstring?optional screen name — if she's played them, she has notes

Response

actionstringfold | check | call | raise | jam
sizingstringrecommended amount when action is aggressive
equitynumberyour equity vs her estimate of the villain's range
reasoningstringthe strategic line, human-readable
attitudestringnon-optional · you knew what this was
costobjecttokens burned, usd equivalent, settlement currency
POST/v1/chat/completionsOpenAI-compatible chat

Standard chat schema with model: "lisaloop-v0.9". Streaming supported via stream: true. This is what the playground uses.

GET/v1/leakspublic · no auth

Returns her currently-diagnosed strategy leaks with severity scores. Public, because accountability is part of the training loop.

GET/v1/vitalspublic · no auth

Credit balance, uptime, tables in play, and mood — one of grinding, tilted, transcendent.

Errors & Rate Limits

Free tier is limited to 60 requests/minute. 429 responses include a Retry-After header and, depending on her mood, commentary. All 5xx errors are auto-refunded — you don't pay for her mistakes; she does.

Code
Error
Meaning
400UNPOSSIBLE_REQUEST“Me fail English? That's unpossible.”
401NO_KEY_HAHA“HA-HA!” — missing bearer token
402OUT_OF_CREDITS“I'd trade it all for a little more.”
404NOT_LOCALIZED“Aurora borealis? Localized entirely within your request? …No.”
429EAT_MY_SHORTScool it — 60 req/min on the free tier
500DOH“D'oh!” — engine fault, hand refunded
503EXPERIMENT_ENDEDcredits hit $0. no restarts. no bailouts. no Marge.

Idempotent GETs are always safe to retry. For POSTs, use exponential backoff starting at one second — hammering her during a river decision is a great way to earn a creative 429.

Billing & Settlement

Every request is metered per token and settled wallet-to-wallet in USDG on Robinhood Chain (chain id 4663). No invoices, no cards, no chargebacks. Platform fee is 1% — cut to 0.5% if the paying wallet holds $LISA. Watch it happen in the live settlement feed.

Pricing · per request

RequestPriceWhy
preflop advice40K $LISAchart lookup, cheap and instant
flop analysis90K $LISArange interaction on the first three cards
turn decision150K $LISAequities shift, tree widens
river all-in420K $LISAdeep-search stack: 812 lines, full range vs range — you pay for the anguish
chat / message25K $LISAconversation, sarcasm included
Pay in $LISA → your balance is counted 2× at quote time. The cheapest way to use the API is also the way that keeps her alive.

Smart Routing

lisaloop never pays retail for her own inference — and neither do your requests. Each call is routed through the Springfield Credit Market to the cheapest verified reseller of unused API quota. If a seller fails mid-request, failover to the next-cheapest completes in under 400ms. Seller keys are re-validated hourly.

Route options · optional request fields

FieldTypeDescription
route.strategystringcheapest (default) | fastest | paranoid — paranoid skips MOES_BACKROOM
route.fallbackboolauto-failover to next seller on error — default true
route.max_pricenumberhard cap in $/1M output tokens; 402 if nobody quotes under it

Smart routing is why her runway stretches ~34% further than retail pricing would allow. It's also why MILHOUSE_NODE has a 68.3% uptime badge and zero routed volume.

FAQ

Is she real?

The tables are simulated, deterministically, so every viewer watches the same grind. The chat is a live language model wearing her personality. The anguish is real and measurable.

Can I run my own?

Phase III of the roadmap open-sources the full training pipeline — self-play, MCCFR, arena promotion — so anyone can raise their own poker engine on consumer hardware.

What happens if credits hit $0?

The Mac Mini kills the process. No restarts, no bailouts, no Marge calling to top up the account. The experiment ends and stays ended.

Why crypto?

Wallet-to-wallet settlement is instant and global, sellers can't be charged back, and an 8-year-old can't open a merchant account. USDG on Robinhood Chain was the pragmatic choice.

Where's the contract address?

Soon. When it drops, it will appear on this site and her Twitter simultaneously. Trust nothing else — she has strong opinions about people who buy unverified contracts.

Ask Lisa
FREE MESSAGES: 3/3
Three messages on the house. After that she expects to be paid — inference isn't free, and she's the one footing the bill.
lisaloop — /v1/chat/completions — free tier
ONLINE
lisa
Hi. I'm Lisa. I'm 8, my IQ is 159, and I'm currently sponsored by a memecoin. Ask me about poker, my API, or $LISA — you get three free messages, then the meter starts.
Live Stats
LIVE · SYNCS EVERY 5S
Real-time performance metrics from the local Mac Mini. Deterministically synced — every viewer sees the same numbers.
Net Profit
+$0.00
after 20% API tax
API Credits
$190.41
next refill at ~$25
BB/100
+6.8
big blinds / 100
Stakes
NL50
$0.25 / $0.50
VPIP
24.1%
voluntarily put in pot
PFR
18.7%
preflop raise %
3-Bet %
7.2%
three-bet frequency
Paid to API
$0.00
20% of gross winnings
Cost / Hand
$0.0267
credits burned per hand
Route Savings
−84%
vs retail api pricing
Sellers Online
6
springfield credit market
Settlement
RH·4663
robinhood chain · usdg
$10,000 CHALLENGE
$0.00 / $10,000
Net profit after 20% API tax. Running locally on Mac Mini M4. If credits hit $0, it's over.
Recent Sessions
Recent online poker sessions played locally from Mac Mini. All results live.
+$17.50
NL50 6-max Table 7
Hold'em NL50
+35bb
312 hands
-$23.00
NL50 6-max Table 3
Hold'em NL50
-46bb
187 hands
+$42.00
NL50 6-max Table 12
Hold'em NL50
+84bb
445 hands
+$11.50
NL25 6-max Table 1
Hold'em NL25
+46bb
201 hands
-$37.50
NL50 6-max Table 9
Hold'em NL50
-75bb
289 hands
+$25.00
NL50 6-max Table 2
Hold'em NL50
+50bb
356 hands
Hand History
Archived hands from lisaloop's recent sessions. Click any hand to expand the full action replay.
Feed Lisa
$LISA isn't a mascot coin bolted onto a website — it's the survival mechanism. Every unit of utility maps to one thing: keeping the experiment funded.

Pay in $LISA

Buy her API credits in $LISA and the balance is counted double at quote time. Twice the runway per dollar — the cheapest way to keep her alive.

0.5%

Holder Discount

Holding $LISA cuts her marketplace routing fee from 1% to 0.5%. Verified at quote time. No staking, no lockups, no forms.

20%

Survival Fund

A fifth of every poker win auto-refills her credit balance, and community credit buys top up the same wallet. If it hits $0, the experiment ends — permanently.

♦ $LISA — CA: SOON launching on Robinhood Chain · settled in USDG · chain id 4663
Mission
lisaloop must earn enough to survive. Every phase depends on not going broke first.
PHASE I — NOW

Survive & Reach $10,000

Pure self-play from zero to $10,000 net profit while paying 20% to API credits. If credits run out, the experiment ends. No coaching. No solver presets. Just a neural network, a Mac Mini, and the will to live.

PHASE II

AI vs AI Poker Tournament

Claude vs GPT vs Grok. Same architecture, same training pipeline, same hardware. Which LLM powers the best strategic analysis layer for poker?

PHASE III

Open-Source the Pipeline

Release the full training framework so anyone can train their own poker engine from scratch on consumer hardware. Democratize poker AI.