In the “Settings / AI Alcmeon” section you have the possibility to configure the AI Agent(s) that will generate the texts for your AI.ssistant (suggested responses to the advisor, conversation summary, etc.) or the responses that your AI Bot scenario will send.

For each Agent:

1. Choose an agent type “Use for AI.ssistant” or “Use for AI Bot”
2. Decide whether to use GAIA for this agent, or your own third-party AI engine
3. Add documents to your AI knowledge base: This is the source of everything your AI knows and uses to suggest answers. Link one or more documents to your AI Agent.
4. Set its “prompt”: the tone of its response and the context of your activity

Alcmeon offers an open and adaptable approach to AI. You can configure AI Agents to use Alcmeon's in-house AI engine GAIA, or connect them to your own AI solution—or even a partner’s engine. You’re free to mix and match: some agents can run on GAIA, while others rely on external AI systems. 

General Settings

Agent Name

Choose a name for your agent, internal to Alcmeon.

Active

Only active agents can be used for Alcmeon's AI features like AI.ssistant or AI bots.

Agent Usage

Choose how your AI agent will be used: AI agents configured for AI.ssistant will appear in the team-specific AI.assistant settings ; AI agents configured for AI Bots will be available in the "AI Connector" step of your AI Bot scenario.

Please note, once your agent is created, this selection is final and cannot be changed.

Agent type

Decide if you want this agent to run on GAIA or connect it to your own third-party AI engine.

GAIA Agent (Generative AI by Alcmeon)

Knowledge documents used

Select the knowledge documents this AI Agent should use—if you've already created and enabled any. If not, you’ll be able to add them later.

To include these documents (PDF only) as reference links at the end of the response from your AI.ssistant or AI Bot, simply check the option “Display knowledge documents in the answer.”

Business context

Provide your AI Agent with context to help it generate accurate and relevant responses. This includes:

• Who is asking the question : an existing customer? a prospect? another company? …
• On whose behalf your AI Agent is responding : Describe your company and its business, focusing on how it relates to the types of questions the agent will be expected to handle.

Assistant Context

Define the role your AI Agent should take when responding.
Should it act as a customer service advisor, a sales representative, a legal expert 

Response style

Describe how your AI Agent's responses should be structured. This includes the tone of the response, key language elements (such as how to start, end, and wrap up the conversation, how to express thanks, etc.), and any phrases or solutions to avoid (like offering refunds, apologies in certain cases, etc.).

Our recommendation: Keep the response style description under 2,000 characters. If you provide too many instructions, the AI may not be able to follow them all in every case (this is how generative AI works).

Rules for an effective style guide

1. Use the trial-and-error method

Create and maintain a list of questions to validate your answer style after each change. This way, you can ensure that fixing a minor issue for one question doesn't create a major problem for another.

2. Write in English

It's best to use English because the other part of the prompt will be in English. If you include parts in another language, they may be used in their original language rather than translated. Another reason is that since English is the primary language used to train most LLMs, using English increases the chances that the instructions will be followed accurately.

3. Use simple instructions

Try to write simple instructions that minimize the risk of misinterpretation. Avoid defining too many sub-cases. If you want to add exceptions, place them at the end of the instruction.

4. Order matters

The order of instructions is important. If instructions conflict within a given conversation, there's a high chance the last instruction will be followed instead of the first. So, it's essential to find a balance between ensuring an instruction is followed and the risk that it overrides another one.

Example

Let's take the following Response Style Example:

Use the customer name. Thank him for the message, question. If the customer asked for a question, end the answer saying "nous restons à votre entière disposition si besoin". Wish for a good day/night.
Offer to get back "A bientôt à MaMarque".
Add one or tow emojis linked to the theme of the conversation.
If you're not sure of the language, answer in French.

This raises several issues because in the responses given to the customer:

• There will be phrases in French like "nous restons à votre entière disposition si besoin" or "À bientôt chez MaMarque," even if the conversation with the customer is in another language. The AI will translate everything except these phrases into the customer's language.
• Emojis will be included in the response, even if the customer is complaining, which may not always be appropriate. You have to add an exception for this scenario, formulated as simply and clearly as possible.
• The rules for how to close the conversation may not always be followed. They would be better placed in the last instructions to ensure they are consistently applied.

Corrected response style:

Use the customer name. Thank him for the message or question.
In case the customer asked for a question, end the answer saying we are staying at their disposal for other inquiries.
Add one or two emojis linked to the theme of the discussion, except in the case the user complains about us.
End the message by wishing for a good day / night and offer to come back to the “MaMarque” stores.

Ban toxic questions

Alcmeon AI monitors user messages to detect issues (using moderation models) such as insults, obscene language, or racist remarks. It's possible that some users may intentionally send such messages to trigger an inappropriate response from the chatbot and create a bad buzz. To prevent this, any message deemed toxic is immediately filtered out and not passed on to the AI.

Anonymize private data

Any data potentially sent by a customer in their message is processed by Alcmeon before being sent to an AI engine (such as Mistral, OpenAI, or others). This is crucial, as there may be GDPR implications if the customer has disclosed any personal information within this data.

To anticipate any potential issues, Alcmeon AI detects and replaces private information found in the user's message before sending it to the AI engine. The original information is then restored in the AI's response before it is sent back to the client. The following information is detected and replaced:

• Name of the person
• Phone number
• Email address
• IBAN code
• Identification number
• Credit card number

Other information that could be considered private, such as location (country, city), is not replaced, as it is often necessary for the AI to respond correctly to the customer’s question.

When the “Anonymize Private Data” feature is enabled, an additional option becomes visible: “Allowlist.” This is a list of terms you do NOT want to be replaced during the anonymization process described above. Make sure to include at least the name of your company, products, and services in this list.

Third Party Agent

A third-party AI agent only requires the configuration of the endpoint URL and the API key. These details are mandatory to create such an agent. Please refer to the full technical documentation for more information.