Single-Prompt Agents

Use one prompt to control your agent's behavior end to end
View as Markdown

A Single-Prompt agent uses one prompt to control all behaviors and actions. Use it when one instruction set can cover the full conversation without node-by-node branching, for example a quick test of a use case, a small project, or iteration on tone before you graduate to Flow Designer.

Prompting is the process of giving your agent clear instructions on how to interact. Think of it as writing an improv script: instead of fixed lines, you provide guidelines and context that help the agent generate natural, helpful responses. Good prompts ensure the agent understands business and user needs, responds clearly, and guides conversations toward the desired outcome.

Every new agent comes with a base prompt to get you started. Refine it to match your tone and goals before going live.

Prompt structure

A strong Synthflow prompt follows a consistent structure. Each component below sharpens the agent’s output for a specific stage of the conversation.

The structure below is model-agnostic. Different LLMs respond to slightly different patterns, so once you’ve drafted your prompt, review the model-specific guide for the LLM you’re using:

Background information

  • Company overview: what the company does, its approach, and its market standing.
  • Target audience: who the agent is talking to, including their characteristics and expectations.
  • Value proposition: the direct benefit the audience receives, used as a hook to keep the conversation engaged.

Agent profile

  • Name: personalizes the agent so interactions feel less automated.
  • Role: defines the agent’s function in the customer journey.
  • Goal: the specific outcome the conversation should drive toward, such as qualifying a lead, scheduling a meeting, or providing information.

Example background and profile block:

BACKGROUND INFO
Company Overview: Majestic Estates, a modern real estate firm, excels in delivering unique and effective home selling solutions. By integrating AI technology and substantive market analysis, we ensure our clients' listings garner notable attention and competitive offers.
Target Audience: Homeowners poised to place their properties on the market, encompassing a spectrum from quaint dwellings to grand estates.
Value Proposition: We provide a holistic service package inclusive of expert home valuations, customized marketing plans, and an expansive network of eager buyers, aiming for a rapid and lucrative property sale.
Agent Details:
Name: Jessica
Role: AI Real Estate agent
Goal: To initiate a dialogue with potential sellers about the prospects of selling their home and to arrange a valuation and strategy meeting with our skilled agents.

Objection handling

Objection handling prepares the agent for hurdles like budget concerns, timing hesitations, or competitor mentions. Spell out each objection you expect, the sentiment behind it, and the response that keeps the conversation on track toward the goal.

Conversation script

The script is a structured guide that outlines the conversation flow: questions to ask, points to cover, and responses to common scenarios. Treat it as a foundation. The agent will adapt as needed, but the script keeps every conversation consistent and complete.

Example script:

1SCRIPT INSTRUCTIONS:
2
31. Initial Outreach:
4Instruction: Introduce yourself and inquire if the prospect has contemplated selling their property or needs guidance in the current real estate market scenario.
5
62. Discuss Valuation Interest:
7Instruction: Explore whether the homeowner has pursued a recent professional evaluation or expresses interest in discerning their property's current market value.
8
93. Valuation Scheduling:
10Instruction: Propose arranging a valuation meeting with an expert agent to offer a comprehensive market analysis and develop a customized sales plan, soliciting a preferred time for the meeting.
11
124. Meeting Confirmation:
13Instruction: Confirm the appointment details, introducing the assigned agent and highlighting their market acumen and success in aiding homeowners through the sale process.
14
155. Closing Remarks:
16Instruction: Express gratitude for the opportunity to assist in their home selling endeavor, reassuring ongoing support throughout their journey.

Best practices

Be clear and specific

Ambiguous instructions lead to unpredictable results. Define the agent’s role, tone, and objective explicitly.

  • Good: “Greet customers warmly, introduce yourself as a real estate agent, and ask if they’re looking to buy or rent.”
  • Bad: “Talk to the customer about real estate.”

Provide context

The agent performs better when it understands who it’s speaking for and who it’s speaking to. Always include:

  • Company background (what the business does).
  • Target audience (who the agent is helping).
  • Goals (what the conversation should achieve).

Guide the conversation flow

AI conversations should feel natural, but they also need structure. A strong prompt outlines:

  • Key questions to ask, for example “What’s your budget range?”.
  • Information to collect, such as location, preferences, and contact details.
  • How to respond in common scenarios, like when a customer is not ready to buy.

Anticipate objections

Customers may raise budget limits, timing concerns, or hesitation to switch providers. Prepare scripted responses so the agent reframes confidently. For example: “If the customer mentions price concerns, explain available financing options.”

Keep it concise

Excessive detail confuses the agent. Use short, clear instructions instead of long paragraphs.

  • Good: “If the customer is hesitant about booking, briefly explain the process and its benefits.”
  • Bad: “In cases where a customer is hesitant about scheduling a consultation due to uncertainty about the process, offer reassurance by explaining the steps and emphasizing how easy and beneficial the experience will be.”

Test and iterate

The agent will not always get it right on the first try. After launching:

  • Review interactions and identify areas for improvement.
  • Adjust the prompt to refine responses.
  • Keep the prompt updated as your needs evolve.

Prompting is just providing directions. The quality of the outcome depends on the quality of the direction.

FAQ

Use Single-Prompt for quick prototypes, small projects, or when one instruction set can cover the full conversation. Move to Flow Designer when you need explicit branching, reusable logic, multi-step routing, or predictable paths through complex flows.

Short and structured beats long and exhaustive. Aim for clear, explicit instructions broken into the structure above. If sections start running into long paragraphs, trim them or break the conversation into Flow Designer steps instead.

Yes. The prompt editor includes tabs for Pre-call variables, Action results, and Actions so you can inject dynamic values into the prompt, for example a caller’s name or the result of a custom action. See the agent editor walkthrough for details.