Investment work can start from the inbox.
Users can send or reply to analysis requests by email, with attachments, share links, follow-up context, and generated outputs tied back to the platform.
Platform Highlights
Institutional work intake, execution, review, and improvement.
The platform is more than research generation. Email workflows, report modes, dynamic work planning, Copilot actions, self-improvement, and AI Investor learning cycles turn repeated use into a stronger operating system.
Platform loops
Compounding system
Differentiation
Work is routed, remembered, and improved.
Email ModeSend work in, receive governed resultsInbox to analysis loop
Report modesQuick, Targeted, Deep, Forum, DesignRight depth for the job
LearningFeedback, diagnostics, AI Investor cyclesSystem improves
Email Modeforward, reply, or request analysis through email while preserving share links, attachments, and follow-up context
Dynamic planswork plans route questions through bounded tools, skills, models, report modes, and vertical-specific evidence profiles
Learning loopsfeedback, diagnostics, AI Investor lessons, and improvement cycles make repeated platform work compound
IntakeWork can start from the app, Copilot, email, uploaded documents, or saved reports without losing context.
ExecutionReport modes and vertical profiles route the job through the right tools, budgets, and model depth.
ImprovementFeedback and diagnostics turn reliability gaps into labeled improvement work instead of untracked complaints.
Platform loop
Every entry point feeds the same operating memory.
App runs, Copilot actions, inbound email, saved reports, diagnostics, and user feedback all connect to the planning and improvement loop.
01 / IntakeApp, Copilot, email, uploads, saved reports, and share linksEvery entry point is normalized into the same governed work queue.
02 / PlanMode, model, source stack, vertical profile, and budgetPlanning chooses depth and tools before the platform spends work.
03 / OperateResearch, monitors, portfolio reviews, strategy runs, and exportsExecution produces artifacts that remain connected to the originating context.
04 / ImproveDiagnostics, feedback, lessons, PRs, and reliability workPlatform feedback becomes visible reliability work rather than hidden drift.
Configured to your office
Your house lens and source stack sit beneath every workflow.
Configure how Starwood AI reasons and what it is allowed to know once. The platform can then route the right philosophy, model, source, document, or dataset into each analysis without rebuilding the stack for every request.
01 / ReasoningInvestment philosophy
Use a built-in framework or encode your own principles, heuristics, risk rules, and IC process.
Judgment layer02 / ComputeBring your own model key
Connect supported AI providers at the account level. Credentials are encrypted and never displayed again after saving.
User key first03 / EvidenceBring your own data keys
Connect supported market, economic, company, property, news, research, and specialist data providers.
Permissioned routing04 / ExtendIntegrate the sources unique to you
Customer-specific internal systems and third-party sources can be connected during onboarding—without creating another research silo.
No fixed source ceilingPlatform sourcesavailable
Your API keysencrypted
Private documents + Data Vaultaccount-scoped
Internal + third-party connectorsextensible
Starwood AI routingThe permitted context follows the decision.
AI Analyst · Monitors · Portfolio · AI Investor
Account Settings is the control surface for philosophies, analysis style, and supported provider credentials.
See the setup guide →Differentiators
Key platform advantages that are hard to copy.
These are the operating primitives that make Starwood AI feel less like a chat app and more like an investment intelligence system.
Depth is selected for the decision.
Quick, Targeted Research, Deep Research, Forum, Design, peer review, counter-thesis, IC simulation, and review-update flows support different kinds of work.
Reliability work is part of the product loop.
User feedback and diagnostics can become labeled self-improvement findings, todos, and reviewable pull requests instead of disappearing into chat history.
Strategies retain lessons across cycles.
AI Investor reviews can learn from prior decisions, monitor movement, paper outcomes, feedback, and mandate-specific operating history.
Work plans adapt to the vertical.
Public equities, healthcare, startups, private equity, real estate, macro, crypto, and other verticals use different evidence bundles and time budgets.
The assistant can operate the platform.
Copilot can navigate, open pages, run analysis, create monitors, manage portfolio and AI Investor workflows, and log feedback with context.
See the platform loops