Skip to main content
The Template Playground includes an AI assistant that can help you create templates, explain code, and answer questions about Accord Project.

​
Set up the AI assistant

Before using the AI assistant, you need to configure it with your preferred AI provider.
1

Open AI configuration

Click the AI settings icon (⚙️) in the top-right corner of the AI Assistant panel, or open it from the sidebar.
2

Choose a provider

Select from the available AI providers:
  • OpenAI (GPT models)
  • Anthropic (Claude models)
  • Google (Gemini models)
  • Mistral (Mistral models)
  • OpenRouter (Multiple models)
  • Ollama (Local models)
  • OpenAI Compatible API (Custom endpoints)
3

Enter model name

Type the specific model you want to use:
  • OpenAI: gpt-5, gpt-5-mini
  • Anthropic: claude-opus-4-1-20250805, claude-sonnet-4-5-20250929
  • Google: gemini-3-pro, gemini-2.5-flash
  • Mistral: mistral-large-latest
  • OpenRouter: openai/gpt-5, meta-llama/llama-3.3-70b-instruct
4

Add API key

Enter your API key for the selected provider. This is stored locally in your browser.
For Ollama (local), you don’t need an API key, but you must run:
OLLAMA_ORIGINS="*" ollama serve
5

Save configuration

Click Save Configuration to store your settings.

​
Open the AI assistant

Once configured, you can open the AI assistant:
  • Click the AI chat icon in the sidebar
  • The AI Assistant panel opens on the right side
  • The panel includes a chat interface with message history

​
Ask questions

You can ask the AI assistant general questions about Accord Project, TemplateMark, and Concerto: 
What is TemplateMark?
How do I define optional fields in Concerto?
What's the difference between olist and ulist?

How do I create a conditional clause?
What's the syntax for accessing nested properties?
How do I format dates in templates?

Why isn't my template rendering?
I'm getting a validation error. What does it mean?
How do I fix this syntax error in my Concerto model?

​
Use context from your editors

The AI assistant can access content from your editors to provide contextual help:

​
Select editor context

At the bottom of the AI chat input, you’ll see three context buttons:
  • TemplateMark: Include your template content
  • Concerto: Include your data model
  • Data: Include your JSON data
Click a button to toggle it on or off. Active contexts are highlighted and used when you send messages.
Enable relevant contexts to get more specific help. For example, enable TemplateMark and Concerto when asking about template syntax.

​
Use specialized prompts

The AI assistant includes preset prompts for common tasks:

​
Text to TemplateMark

Convert plain text into a TemplateMark template:
1

Select the prompt

Click the up arrow (^) button next to the send button and select Text to TemplateMark.
2

Paste your text

Enter the plain text of an agreement:
This agreement is between Acme Corp and John Smith. 
The project will be completed by December 31, 2024. 
The total fee is $5,000.
3

Send and apply

The AI generates a TemplateMark template with variables extracted from your text.

​
Create Concerto Model

Generate a Concerto data model based on a description:
1

Select the prompt

Click the up arrow button and select Create Concerto Model.
2

Describe your agreement

Explain the type of agreement and data fields you need:
Create a model for a rental agreement with landlord name, 
tenant name, property address, monthly rent, and lease start/end dates.
3

Review and apply

The AI generates a complete Concerto model with appropriate field types.

​
Explain selected code

You can ask the AI to explain specific parts of your template:
1

Enable code selection menu

Make sure Enable Code Selection Menu is checked in AI settings (it’s on by default).
2

Select code

In any editor, highlight the code you want explained.
3

Click Explain

A menu appears near your selection. Click Explain to open the AI assistant with your selection.
4

View explanation

The AI provides a detailed explanation of what the code does.

​
Get inline suggestions

The AI can provide ghost text suggestions as you type:
1

Enable inline suggestions

In AI settings, check Enable Inline AI Suggestions (enabled by default).
2

Start typing

As you type in any editor, the AI suggests completions based on context.
3

Accept suggestions

Press Tab to accept the suggestion, or keep typing to ignore it.
Inline suggestions work best with fast AI models. Slower models may cause noticeable delays.

​
Apply AI-generated code

When the AI generates code in its responses, it can automatically apply it to your editors:
  • Look for code blocks with Apply buttons
  • Click Apply next to a code block
  • The code is inserted into the corresponding editor (Concerto, TemplateMark, or JSON)
The AI uses special language tags like concertoApply, templatemarkApply, and jsonApply to indicate which editor should receive the code.

​
Advanced settings

Click Advanced Settings in the AI configuration dialog for additional options:

​
Max Tokens

Set the maximum number of tokens the AI can generate in responses. Higher values allow longer responses but may be slower.

​
Show Full Prompts in Chat History

When enabled, the chat history shows the complete prompt sent to the AI, including system messages and context. This is useful for debugging or understanding how the AI processes your requests.

​
Enable Code Selection Menu

Toggle the “Explain” menu that appears when you select code in editors.

​
Enable Inline AI Suggestions

Toggle ghost text suggestions as you type.

​
Manage your conversation

​
Reset chat

Click the reset icon (↻) in the AI Assistant header to clear the conversation history and start fresh.

​
Stop generation

If the AI is generating a long response, click Stop to halt it mid-stream.

​
Close the panel

Click the X button to close the AI Assistant panel. Your conversation history is preserved until you reset it.

​
Best practices

Provide clear, detailed questions:
  • ❌ “My template doesn’t work”
  • ✅ “My template shows an error when I try to iterate through a list using #olist. What’s the correct syntax?”
Enable editor context when asking about your specific template, but disable it for general questions to save tokens and get faster responses.
If the AI’s first response isn’t perfect:
  • Ask follow-up questions
  • Request modifications
  • Ask for explanations of specific parts
Always review AI-generated code before applying it:
  • Check that field names match your model
  • Verify the syntax is correct
  • Test in the preview before using in production

​
Privacy and security

Your API keys and template content are sent to the AI provider you select. Be cautious about:
  • Using sensitive or confidential data in templates when AI is enabled
  • Sharing API keys or configuration
  • Understanding your AI provider’s data policies
  • API keys are stored locally in your browser (localStorage)
  • Template content is only sent when you have editor context enabled
  • No data is sent to Accord Project servers

​
Reset AI configuration

To completely reset your AI settings:
  1. Open AI Configuration
  2. Click Reset Configuration at the bottom
  3. Confirm the reset
  4. All settings, API keys, and preferences are cleared

​
Next steps

Creating templates

Learn template creation fundamentals

Exporting documents

Export your templates as PDF or HTML

Build docs developers (and LLMs) love

Get started for freeTalk to us