Agents
Create durable AI agents that automate investment research tasks on a schedule or on-demand
Open in AppOverview
AIIME Agents are durable, autonomous AI workflows that run investment research tasks on your behalf — either on a schedule or on-demand. Each agent combines natural-language instructions with the platform's full suite of 80+ financial data tools to execute multi-step research processes without manual intervention. Agents are powered by Temporal for reliable, fault-tolerant execution, so even long-running tasks complete successfully.
Core Concepts
| Concept | Description |
|---|---|
| Agent | A saved configuration with a title, instructions, selected tools, trigger type, and notification preferences |
| Run | A single execution of an agent — records output, steps taken, tools called, duration, and status |
| Memory | A persistent scratchpad unique to each agent — the agent can read and write to this across runs to build context over time |
| Tool | Any of the 80+ data tools available on the platform (FMP, Neptune rankings, technical analysis, options flow, web search, etc.) |
| Trigger | How the agent starts: manually (on-demand) or on a cron schedule |
Creating an Agent
Navigate to AIIME Agents from the sidebar and click “New Agent.” Fill in the following fields:
- ●Title — a short name to identify the agent (e.g., "Morning Market Brief" or "Weekly Sector Rotation Scan")
- ●Instructions — natural-language directions telling the agent what to do. Use the / key to auto-suggest tools inline. Be specific: describe the analysis, output format, and any tickers or criteria.
- ●Tools — select the data tools the agent can use. Tools are grouped by category (Market Data, Technical Analysis, Fundamentals, etc.). You can also let the / shortcut auto-select tools from your instructions.
- ●Trigger Type — choose Manual (run on-demand) or Scheduled. Scheduled agents use cron expressions with built-in presets like "Every weekday at market open" or "Every Monday at 8 AM ET."
- ●Notifications — choose whether to be notified on success, failure, or both. Notifications can be sent via email, AIIME chat, or both.
Tip
Trigger Types
| Trigger | When It Runs | Best For |
|---|---|---|
| Manual | When you click "Run Now" on the agent card | Ad-hoc research, testing new instructions, one-off deep dives |
| Scheduled | Automatically at the configured cron time | Daily market briefs, weekly scans, recurring reports |
Schedule Presets
- ●Every weekday at 8 AM ET — morning prep before market open
- ●Every weekday at market open (9:30 AM ET) — real-time open analysis
- ●Every weekday at market close (4 PM ET) — end-of-day summary
- ●Every Monday at 8 AM ET — weekly kickoff scan
- ●Every hour during market hours — intraday monitoring
- ●Daily at 7 AM ET — early morning brief including pre-market data
Note
Agent Memory
Each agent has its own persistent memory — a scratchpad that carries context between runs. The agent can read its memory at the start of each run and write updates at the end. This allows agents to track evolving situations, build watch lists over time, or refine their analysis across multiple executions.
- ●Memory is stored per-agent and persists across all runs
- ●The agent automatically reads its memory at the start of each execution
- ●To write to memory, the agent outputs a ```memory block in its response
- ●You can also manually view and edit an agent's memory from the Memory tab on its detail page
- ●Memory is versioned so you can track changes over time
Tip
Run History & Auditability
Every agent run is recorded with full auditability. Click an agent to see its run history, then drill into any run for step-by-step details.
| Field | What It Shows |
|---|---|
| Status | Whether the run completed successfully, failed, is running, or was cancelled |
| Duration | Total wall-clock time for the run |
| Output | The full text output produced by the agent |
| Summary | A brief auto-generated summary of findings |
| Steps | Every tool call made, reasoning step, and memory write — with inputs, outputs, and timing for each |
| Error | If the run failed, the error message explaining what went wrong |
Notifications
Agents can notify you when they finish a run. Configure notifications when creating or editing an agent:
| Channel | How It Works |
|---|---|
| Sends the agent output as a formatted email to your account email address | |
| AIIME Chat | Delivers the agent output as a new message in a dedicated AIIME chat session |
| Both | Sends via both email and AIIME chat simultaneously |
Note
How to Use
- ●Navigate to AIIME Agents from the sidebar
- ●Click "New Agent" to create your first agent
- ●Write clear instructions describing the research task
- ●Select the tools your agent will need (or use / in instructions to auto-select)
- ●Choose manual or scheduled trigger and set a cron schedule if needed
- ●Configure notification preferences
- ●Save the agent and click "Run Now" to test it immediately
- ●Review the run output and steps to verify the agent works as expected
- ●Adjust instructions, tools, or memory as needed and re-run
- ●Once satisfied, set the trigger to scheduled for automated execution
Example Agent Ideas
| Agent Name | Instructions Summary | Schedule |
|---|---|---|
| Morning Market Brief | Summarize S&P 500, Nasdaq, top movers, and key macro events for today | Weekdays at 8 AM ET |
| Weekly Sector Rotation | Analyze all sector ETFs by momentum and relative strength, rank the top 3 and flag any regime changes | Every Monday at 8 AM ET |
| Earnings Watch | Check this week's earnings calendar, pull fundamentals for each reporting stock, and flag any with strong composite scores | Weekdays at 7 AM ET |
| Options Flow Scanner | Pull unusual options activity, filter for large call sweeps on stocks with bullish technicals | Every hour during market hours |
| Portfolio Risk Check | Analyze the technical signals and drawdown risk for AAPL, NVDA, MSFT, GOOGL, AMZN | Weekdays at market close |
Combining with Other Tools
- ●Use AIIME Chat to ask follow-up questions about an agent's output using the Agent Selector
- ●Set up one agent for screening and another for deep-dive analysis on the screened results
- ●Use agent memory to build evolving watch lists that feed into your Screener analysis
- ●Reference ALEEP Signals in your agent instructions for systematic signal confirmation
This platform provides data and analysis tools for educational and informational purposes only. Nothing on this platform constitutes financial advice, investment recommendations, or solicitation to buy or sell any securities. Always conduct your own research and consult with a qualified financial advisor before making investment decisions.