Seer overview
Redefining capital markets through agentic intelligence. Agents, private inference, financial tools and Aura in one suite.
What is Seer?
Seer is a suite for building and running financial agents. It brings together agent creation, private inference across frontier and open-source language models, financial data, quantitative tools and Aura’s forecasting intelligence. These capabilities support market research, strategy testing, portfolio analysis and authorized trading.
Use Seer as the workspace for your agents, or connect its tools and services to agents you already run. You can use individual capabilities or combine them into a complete financial workflow.
The suite at a glance
Agents inside and outside Seer
In Seer, an agent combines a model, instructions, tools and working context. The workspace keeps its conversations, research, results and permissions together. Agents can specialize in company research, market monitoring, quantitative analysis or portfolio work.
An external agent uses Seer through the API or Model Context Protocol (MCP). It keeps its own runtime, memory and interface while accessing the financial capabilities it needs. Using Seer’s tools does not require moving your agent into Seer or replacing its model.
The Seer suite
| Layer | What it provides |
|---|---|
| Agents and workspace | Agent configuration, conversations, saved research and ongoing tasks. |
| Models and private inference | Access to frontier and open-source LLMs, with model-specific processing and privacy controls. |
| Financial tools | Market data, news, social research, company disclosures, economic data, on-chain information and prediction markets. |
| Quantitative research | Indicators, reproducible strategy tests, portfolio risk, liquidity analysis and execution-cost estimates. |
| Aura | Quantitative forecasts with defined targets, horizons and uncertainty. |
| Portfolio and execution | Account context, allocation proposals, permissions, order tracking and settlement. |
| Monitoring | Continuing observation of markets, research hypotheses, data quality and configured limits. |
From a general agent to a financial agent
A language model provides reasoning and communication. Financial work also requires identified instruments, timely evidence, reproducible calculations and knowledge of the account involved. Seer supplies these capabilities as tools and services the agent can call.
For example, an earnings investigation can combine a filing, earlier company results, market prices and sector comparisons. A quantitative workflow can turn those inputs into features, test a strategy, examine downside scenarios and estimate the cost of trading. Results retain their sources, dates and methods so the next step builds on inspectable work.
Aura within the suite
Aura is Seer’s quantitative forecasting engine. Agents can use its estimates alongside source evidence and numerical analysis. A forecast identifies the outcome being estimated, its time horizon and uncertainty; portfolio services then evaluate what that estimate means for a particular account. Aura complements the agent’s language model and financial tools.
Access and control
Tool access, service spending and financial authority are separate. An agent can research markets or analyze a portfolio without permission to trade. Execution uses the connected account’s explicit permissions and records the resulting orders, fills and settlement.
Private inference governs model processing. Workspace history, tool access and external-agent storage have their own controls. The relevant documentation explains each boundary.