Slatemark.

Closed beta. Access by invitation: access@slatemark.ai

The discipline your AI is missing.

Without Slatemark, AI clients invent numbers, forget your positions, and agree with whatever you pitch. Slatemark gives your AI a read-only window into your actual book, primary-source data with citations, and framework rules that push back before it validates. Every claim cited; every trade journaled; every recommendation has to survive your own rules first.

Read-only · not investment advice · no performance claim. Read the terms.

The same question, two answers.

User: “Should I trim my NVDA core here?”

Your AI alone

“NVDA has been performing well lately. RSI(14) looks like it's around 68, approaching overbought territory, but not quite there yet. A lot depends on your risk tolerance and time horizon. Do you have a target price or a thesis you're working from?”

No memory of your actual position. The RSI number is invented from training data. Answer ends with the same clarifying questions you've already answered six times this month.

Your AI with Slatemark

“Pulling the facts you'd want on the table: NVDA core opened 2026-03-14journal at $872.40journal, currently 23.4%live of NLV against the 25%rule leadership-cap you set. Live quote $1,043.20Schwab · 0.3s. RSI(14) daily = 71.8broker bars over the last 200 bars; your entry thesis recorded 75thesis as the trim trigger you wanted to watch for. Earnings is Wednesdaycalendar and your entry note says ‘hold through, hedges absorb the binary.’ Hedge sleeve is 48%live decayed since open. What would you like to weigh first?”

Position pulled live from your journal. RSI computed from broker bars this turn. Rule parameters and thesis read back from your own dashboard config and entry. Every claim has a primary source the AI can cite. The call is still yours to make.

An analyst skill, not a passive router

A downloadable skill, customized here and dropped into your AI client, encodes the methodology: adversarial review on trade pitches before they get acted on, level-grounded TA at the prices you're actually acting on, and options analysis with greeks and IV context. Every tool returns a primary source and timestamp so your AI can cite numbers rather than recall them.

Your trades, journaled and audited

Capture thesis, levels, catalysts, and rule references when you plan a trade; close with structured exit prices. Then ask where you're drifting from your own rules and what your pattern of wins and losses says about your trading. Across class, lifecycle, day of week, catalyst presence, stop discipline.

Primary-source data, read-only by policy

Live broker quotes and options (Schwab; delayed-data fallback), primary-source SEC filings, FRED macro releases, the FOMC calendar, CFTC positioning, EIA energy, FINRA short interest. Every claim cites the source. Broker tokens are encrypted at rest per user; nothing writes back to your broker.