Original Article
Read the full article on X.
Written by Pete (@peteferr) on July 24, 2026 — outlining the complete vision for Teresa on Robinhood.
Pete (@peteferr) · Jul 24, 2026
Why We Built Teresa
on Robinhood.
"Own the layer where research becomes a decision, and let the settlement layer commoditize underneath you."
TERESA NETWORK
AI Trading Infrastructure · Robinhood Chain
1. The Market Structure Read
The Asymmetry Thesis.
We spent weeks studying where tokenized equities actually stand before writing a single line of code.
01
Robinhood Made Equities a Chain
Robinhood moved equities on-chain in 2025: an Arbitrum Orbit L2 with USDG settlement, open enough for venues like Rialto & Ostium to route to it.
02
The Missing Interface Layer
No native terminal existed for tokenized equities — no charting, analytics, options data, filings, or sentiment. Traders were bridging assets in one tab and reading news in another.
03
Bloomberg Outlived Every Exchange
History is clear: venues compete their margins away while the interface layer retains pricing power. Bloomberg outlived every exchange it connected to.
04
Where Research Becomes a Decision
Own the layer where research becomes a decision, and let the settlement layer commoditize underneath you. That asymmetry was our thesis.
12-Seat AI Multi-Agent Desk
Structured like an actual trading floor.
Research must end in a position, not a summary. Single models hallucinate numbers — multi-agent debate argues against computed context.
Analyst Seats
Dedicated agents reading tape structure, news feeds, market sentiment, and fundamental earnings data in real time.
Bull & Bear Debaters
Bull and Bear agents running multi-round rebuttals against each other's specific claims, citing computed context instead of vibes.
3 Risk Seats
Reprices setups across 3 risk budgets, setting daily sigma stops, VaR, and CVaR derived directly from realized distributions.
Portfolio Manager (PM)
Commits to a direction-adjusted trade call with target levels and a 21-day probability-weighted scenario tree (p05 to p95).
The 3 Core Targets
What we held ourselves to.
From our market read, we established three strict product commitments and built everything around them.
Target 1: Research ends in a position, not a summary
12-seat floor debating computed context — trend regime, VIX, cross-asset relative strength against SPY, and 21-day projection cones.
Target 2: Real non-custodial execution
Perps run long and short with leverage on Ostium. Desk levels draw on charts with 1-click tickets closing the distance to position.
Target 3: Social trading floor
Wallet-linked profiles, verified handle proofs, invite-only groups, and a live trading feed. Trading is social or it's lonely churn.
Teresa Stack
Robinhood Orbit L2
Research & Debate
12-Seat AI Desk
Computed Context
1-Click Ticket
Non-Custodial Terminal
Ostium Perps
Social Feed
USDG Settlement
Robinhood Chain
Tokenized Stocks
2. The Economic Layer
Deliberately boring tokenomics.
Supply reduction is a linear function of research demand. Execution fees recycle back into liquidity and tokenized equities.
Credit Burn Engine
The desk runs on credits bought with the token. Spent tokens burn, making supply reduction a linear function of research demand and nothing else.
spent_tokens.burn() // linear burn
Liquidity Recycling
A portion of execution fees recycles to deepen the token's liquidity pool, so market depth scales with trading volume instead of rented incentives.
recycle_fee → LP_depth
Stock Airdrop Pool
Fees market-buy tokenized stocks on Robinhood Chain, which are airdropped to users who earn allocation spots through activity. Real equities funded by real fees.
fees → marketBuy(TokenizedStocks) → Airdrop