Transforms a_recommend

Recommend

LLM-reasoned recommendations from a candidate list with a greeting line.

What does it do?

Recommends products or content based on user preferences.

How does it work?

Compares the user's profile or query against your database and uses AI to find the most relevant items.

Why use it?

Boosts your sales (upselling/cross-selling) by suggesting the perfect product to the customer without them having to search for it.

Inputs

name	type
`text`	string
`top_hits`	array

Outputs

name	type
`greeting`	string
`recommendations`	array

Branches

Fixed branches

out — Output

Configuration

candidates_field string required default: "top_hits"

query_field string required default: "text"

limit integer required default: 3

model string optional

max_tokens integer required default: 800

temperature number required default: 0.3

display_fields array<string> required

reasoning_prompt string optional

greeting_style "casual" | "formal" | "sales" | "consultative" required default: "consultative"

memory_key string optional

output_field string required default: "recommendations"

emit_blocks boolean required default: false

block_actions array<string> optional

Default configuration

{
  "candidates_field": "top_hits",
  "query_field": "text",
  "limit": 3,
  "model": "deepseek/deepseek-v4-flash",
  "max_tokens": 800,
  "temperature": 0.3,
  "display_fields": [
    "name",
    "price"
  ],
  "greeting_style": "consultative",
  "output_field": "recommendations"
}

Emitted events

This node does not publish events on the pubsub channel.