Interacting well with 2Rai speeds up useful outcomes: better summaries, more applicable playbooks, and answers that need less human validation. Below you’ll find simple techniques and ready‑to‑use examples: role prompting, delimiters, step‑by‑step instructions, one/few‑shot prompting, and how to ask 2Rai to improve its own responses.
1. Role prompting — put the assistant in the right role
What it is: Explicitly tell 2Rai which “role” to assume so it adapts tone, technical level and output format.
Why use it: You avoid follow‑up clarifications and get responses with the right focus from the first turn.
Examples:
- For an executive version:
“You are an executive summary writer. Summarize the following findings in 4 bullets clearly for C‑level, avoiding technical jargon.”
- For a technical analysis:
“You are a senior security analyst. Explain the technical impact of this finding and list remediation steps prioritized by risk.”
- For a technical analysis:
2. Delimiters — clearly define the content and its context
What it is: Use markers to separate instructions from the material you upload or the text you want analyzed.
Why use it: Prevents the model from mixing instructions with data and ensures it processes only the intended content.
Useful delimiter patterns (use any of these plain text markers):
- START_DOC / END_DOC
- <<<DOCUMENT>>> … <<<END>>>
- —-DOCUMENT—- (separator line)
Example:
“Instructions: Summarize the content between the delimiters. <<<DOCUMENT>>> [Paste the report text here] <<<END>>>
Summary in 5 bullets.”
3. Step‑by‑step instructions — guide the format and process of the output
What it is: Explicitly request a structured response process (for example: context → findings → recommendations → verification criteria).
Why use it: You get actionable outputs ready to use (checklists, playbooks).
Example prompt:
“Read the document between delimiters. Then produce:
1) Short context (1 sentence).
2) Top 3 findings with evidence (each with source/page).
3) Remediation steps as a numbered checklist.
4) Estimated priority (High/Medium/Low).
<<<DOCUMENT>>> … <<<END>>>”
4. One‑shot and few‑shot prompting — show the example you want it to follow
What it is: Provide one (one‑shot) or several (few‑shot) input/output examples so 2Rai copies the format and style.
Why use it: Useful when you need a specific format (e.g., a report template) or a particular tone.
- One‑shot example:
“Example input: [short excerpt]. Example output: [one paragraph summary]. Now process the document below and return a similar one‑paragraph summary. <<<DOCUMENT>>> … <<<END>>>”
- Few‑shot example (two examples):
“Example 1 input: … Example 1 output: …
Example 2 input: … Example 2 output: …
Now produce the output for the following document in the same format: <<<DOCUMENT>>> … <<<END>>>”
- Few‑shot example (two examples):
5. Ask 2Rai to improve / iterate — use the assistant to refine its own answers
What it is: Ask 2Rai to review and improve its previous output, or to ask clarifying questions before responding.
Why use it: Reduces back‑and‑forth and improves accuracy without making the user craft ever‑more complex prompts.
Practical patterns:
- Self‑review: “Evaluate your previous answer for clarity and completeness. Provide an improved version with one paragraph of rationale for each change.”
- Clarifying questions first: “Before answering, ask up to 3 clarifying questions that will help you produce an actionable remediation plan.”