Chat

Chat helps you work with AI using your project context, cited sources, attachments, and retrieval controls.

What Chat Is For

Chat is your AI working interface inside Engenerate. It is designed to help you explore and interrogate your project documents, draft content grounded in source material, compare information across multiple references, and move from raw material to useful outputs faster.

Chat operates with:

• Retrieval from your project's processed documents
• Citations that link back to the source
• File and photo attachments for additional context
• A multi-tab interface to keep multiple conversations open
• Configurable retrieval settings for precision control

Chat Tabs

Chats are organized as tabs. Each tab represents an independent conversation.

• New tabs are created by clicking the + button in the chat panel
• Tab names are assigned automatically after the first exchange in each conversation, so you can identify threads without manually naming them
• All previous conversations are preserved in the chat history and can be returned to at any time

This multi-tab design is especially useful when you are working through multiple questions simultaneously, comparing scenarios, or exploring different aspects of a project in parallel.

Models

Engenerate supports two model categories: Base and Premium.

Base Models

Base models are included in your monthly plan allowance. They are the right choice for most everyday chat work.

Model	Best For
Standard	Fast, general-purpose responses; everyday project questions
Advanced	Deeper reasoning and more nuanced responses when the task warrants it

Both base models are capable, well-suited for document-grounded work, and covered by your plan's monthly chat allowance.

Premium Models

Premium models give you access to additional model families from leading providers. Premium usage is always charged against your credit balance on a pay-as-you-go basis with 0% markup—meaning you pay exactly what the model costs, with no hidden surcharges.

Currently available premium providers:

• OpenAI — Advanced GPT-series models
• Anthropic — Claude-series models
• Gemini — Google Gemini-series models

When you select a premium model, an estimated credit cost is shown next to the model name so you can make an informed choice before starting a conversation. Premium models are grayed out when your credit balance is insufficient.

Monthly Base Chat Allowances

Each plan includes a monthly allowance of base model chats that resets at the start of each billing period. For current allowance limits by plan, see the Pricing page.

Base chat allowance applies only to base models. Premium model usage always consumes credits regardless of plan.

Conversation Actions

Each message exchange supports the following actions on assistant responses:

Copy Message

Copy the full text of any response to your clipboard.

Send to Notebook

Save any assistant response directly to an open notebook. The content is inserted into the notebook with its context preserved. This is one of the most efficient ways to build a working draft or report from grounded AI responses.

Citations

When chat retrieves content from your project documents, responses include citations. Citations appear inline with the response text and identify the source document and section.

Clicking a citation opens the referenced document at the relevant page so you can inspect the source in context. This keeps AI-generated answers inspectable and traceable rather than opaque.

Context Selection

By default, chat can search across all processed documents in your project to find relevant context before answering. You can adjust this behavior from the Context panel.

Pin specific documents — Select one or more documents from the context list to give them higher priority in retrieval. Pinned documents are searched with boosted weight, making them more likely to appear in the context provided to the model. This is most useful when you want the conversation to stay anchored to a specific report, drawing set, or reference document.

When no documents are pinned, the system searches broadly across the project. When documents are pinned, retrieval is biased toward those sources while still considering other project content.

The Context panel also includes a Library tab. From this tab you can pin documents from the curated Library — standards, regulations, and reference guides — to include them as additional context in your conversation. Library documents appear alongside your project documents in citations.

Retrieval Settings

The retrieval settings let you fine-tune how the system searches your document library for context. The defaults work well for most users. These controls are most useful when you want tighter or broader grounding behavior.

Setting	What It Does
Document Retrieval	Turn retrieval on or off entirely. When off, the model responds without searching your documents.
Search Scope	Choose whether chat searches the current project only or across all your projects.
Search Mode	Vector search uses semantic similarity to find conceptually relevant content. Hybrid search combines semantic and keyword matching for cases where exact terminology matters.
Top K	Controls how many candidate document chunks are considered. Higher values cast a wider net; lower values are more focused.
Minimum Score	Filters out low-confidence matches. Raising this threshold keeps only the most relevant results.
Exclude Navigation Blocks	When enabled, filters out structural page elements like table of contents entries, headers, and footers that rarely contribute useful context.

For most document-grounded workflows, the default settings are a good starting point. Switch to Hybrid search when your questions rely on specific technical terms, identifiers, or nomenclature that semantic search alone might miss.

Adding Photos and Files

Use the + button in the chat prompt area to attach additional context to a message:

• Photos and images — Attach an image to ask a visual question or include a figure for reference
• Files — Add a supporting document for one-off context without uploading it as a project resource
• CSV files — CSV uploads can be routed directly into the Datasets workflow so the data becomes a reusable project asset

Attachments add context to a single message and are not automatically indexed into your project's knowledge base.