Advanced prompting techniques

How sophisticated can an AI Agent on Public get? This chapter starts with a simple recurring investment and builds on it one instruction at a time, adding a technical indicator, a purchase limit, a prediction-market signal, precise timing, and an options position.

Start with a simple prompt

Begin with a straightforward recurring investment:

Example prompt

Invest $500 in TSLA every Friday.

Add a market condition

To be more selective about when purchases happen, add an instruction:

Example prompt

Only buy if Tesla’s RSI is below 30.

Now the Agent checks a market condition before buying. If Tesla’s RSI is 30.1 on a Friday, there’s no trade. Once an Agent is live, it follows the exact rules you’ve approved. It won’t decide that 30.1 is close enough; it checks the condition, sees it hasn’t been met, and skips that week’s purchase.

You can also limit how often the strategy runs:

Example prompt

Don’t make more than one purchase per month.

The Agent now combines a recurring schedule, a technical indicator, and a limit on how often it can buy.

Use prediction-market signals and precise timing

Agents can also use information from prediction markets to inform trades in other assets:

Example prompt

Only buy if the market-implied probability of Tesla delivering more than 400,000 vehicles next quarter is at least 75%.

You can be more specific about timing, too:

Example prompt

Check these conditions every Friday at 3:50 PM Eastern.

The Agent now combines a schedule, a technical indicator, a prediction-market signal, and a limit on how often it can act.

Add an options step

Finally, an Agent can take a follow-up action after a trade:

Example prompt

After a purchase fills, if I have at least 100 eligible shares of TSLA, sell one covered call against them, targeting a delta of 0.25 and an expiration next month.

The strategy has gone from a simple recurring investment to one that checks multiple market signals, buys a stock, and may open an options position, all from a single prompt.

Build as much detail as your strategy needs

Agents can work with a wide range of triggers, from specific times and prices to technical indicators, market events, and the execution of other orders, and you can combine them into increasingly sophisticated workflows. The goal isn’t to make every Agent as complicated as possible, but to include as much detail as your strategy requires. Nothing goes live until you’ve reviewed and approved your Agent’s instructions.

Key takeaways

  • You can build a sophisticated Agent by adding instructions to a simple prompt one at a time.
  • Agents can combine schedules, technical indicators like RSI, purchase limits, and precise timing.
  • Agents can use prediction-market probabilities to inform trades in other assets.
  • An Agent can take follow-up actions, like selling a covered call after a purchase fills.
  • Active Agents follow your approved rules exactly, and nothing goes live until you review and approve it.

Frequently asked questions

How do I write an advanced prompt for an AI Agent?

Start with a simple instruction, then add conditions one at a time. For example, you can add a technical indicator, a limit on how often the Agent buys, a prediction-market signal, specific timing, or a follow-up action like an options trade. Include as much detail as your strategy requires.

Can AI Agents use technical indicators like RSI?

Yes. For example, you can tell an AI Agent on Public to buy only if a stock’s RSI is below 30. Once live, the Agent checks that condition each time it runs and skips the trade if it isn’t met.

Can AI Agents use prediction market data?

Yes. AI Agents on Public can use information from prediction markets to inform trades in other assets. For example, an Agent can buy a stock only if the market-implied probability of a specific event is at or above a level you set. Keep in mind that prediction market prices reflect what traders are willing to pay, not a guarantee of any outcome.

Can an AI Agent combine stock and options trades?

Yes, if your account is already approved for options trading. For example, after a stock purchase fills, an Agent can sell a covered call against the shares if you hold at least 100 eligible shares, using a target delta and expiration you specify. Options trading involves significant risk and is not suitable for all investors.

Up next
You’ve completed the course
Back to all chapters