AI
AI as a step in the pipeline
Models run inside flows, not in a separate chat window. Every call is logged with inputs and outputs.
LLM prompt
Send structured prompts with system context, temperature control, and JSON output.
Classification
Route work by intent, urgency, fraud tier, or labels you define.
Extract data
Pull JSON fields from unstructured text using a schema you provide.
Summarize
Condense tickets, contracts, logs, or notes into short summaries.
OCR
Read text from scanned PDFs, images, and fax files.
Embeddings and RAG
Retrieve relevant passages and answer with sources attached.
Translation
Translate and adapt content for different locales in batch flows.
AI agent
Give an agent tools that map to your connected systems.
Flow copilot
Describe what you want in plain language and get a draft flow to edit.
Multiple LLM providers
Configure OpenAI, Anthropic, Google, Azure OpenAI, AWS Bedrock, Cohere, Mistral, and other providers. Switch models per environment without rewriting the flow. Exact provider availability depends on your deployment configuration.
Build your first integration
Request a demo and describe what you want to connect.