• +

We run AI teams that research, decide, and trade.

Intent is a systematic trading firm whose analysts and traders are AI agents. Some carry a thesis for months, others work the tape by the second. Every decision passes the same risk gate and lands in the same record. We run our own capital this way today.

Investment decision
PaperLive

From research to a filled order

Analyst team filed the semiconductor briefMon 09:12

Datacenter capex is accelerating and supply stays tight into next year. Sources attached.

Investor agent sized the allocationWed 14:05

Increase the position to 4% of the portfolio. The thesis fails if datacenter orders roll over.

Risk gate passed every checkThu 10:30

SizingLiquidityExposure

Execution filled and filed the orderThu 10:31

Filled at 176.20 at the venue. Brief, thesis, checks, and fill saved to the record.

01 · The system

From research to a filled order.

We run two tempos on one pipeline. A thesis that takes months and a setup that takes seconds meet the same controls, book to the same venue, and leave the same paper trail.

Research & invest

days to months

Our analysts divide a sector and challenge each other's briefs. The investor agent allocates on the evidence.

fundamentals · filings · earnings · industry data

Trade the market

24/7

Our market agents read the tape, enter systematically, and rest their stops at the venue.

order flow · volume profile · market structure

Every decision, whichever engine it came from

01

Risk gate

We size every order from the mandate's risk budget and attach a stop.

02

Venue

The stop rests at the venue, not inside our code.

03

Record

The brief, thesis, checks, fill, and post-mortem all sit under the agent that wrote them.

We set the limits, approve what needs approving, and one switch stops everything. The agents work inside that.

02 · Method

We only trade what we can test.

We work from primary data: the tape for trading, the filings for investing. We replay every agent against history before it touches capital.

For the market engine

Order flow

Who is hitting bids and lifting offers, at what size.

Volume profile

Where the market transacted, and where it did not.

Market structure

Levels, trend, and liquidity, read from price.

For the research engine

Filings and earnings

Reports, results, and guidance, read in full with sources kept.

Industry and supply chains

Sector data and the links between companies.

News and the calendar

Scheduled events and headlines, weighed as context.

How we prove an agent

before capital
  1. 1

    Replay against history

    Bar by bar on data it has never seen. No lookahead, no shortcuts.

  2. 2

    Trade on paper

    Whatever survives the replay then trades simulated capital under production conditions.

  3. 3

    Promote to live

    Only validated agents reach real capital, and they reach ours first.

03 · The technology

We build the machine ourselves.

We are an engineering firm first. The platform our agents run on is ours, so we can validate it, audit it, and stand behind it.

01

Agent orchestration

Our agents run as durable teams with versioned instructions and full run histories.

02

Shared agent workspace

Agents hand work through one file system that a person can also open.

03

Replay & backtest engine

The same code runs on historical and live markets. What we test is what trades.

04

Limits in code

Risk rules sit between the agent and the venue. Code enforces them, not policy.

04 · Strategies

Our strategies trade our capital first.

We run systematic strategies across equities, indices, crypto, and FX. What they trade and how they decide stays off this page on purpose.

What we run

today
Asset classes
equities · indices · crypto · FX
Approach
systematic, risk-gated
Holding period
intraday to multi-month
Coverage
24/7

For qualified investors

We share mandates, the decision process, and the live track record privately.

Request the deck

Work with us

Every trade carries the research behind it.

Request access and we will walk you through the mandate, the process, and the record in private.

Request access

How we start

01

Agree the mandate

We fix markets, risk limits, and horizon in writing, then compile them into the risk gate.

02

Start on paper

Your mandate runs on simulated capital under production conditions first.

03

Go live

Track the portfolio, read the trail behind any trade, and widen the mandate as trust builds.

Intent — AI teams that research, decide, and trade