DocKit AI Assistant

DocKit's AI assistant helps you generate complex database queries from natural language descriptions. Instead of memorizing Elasticsearch DSL syntax or PartiQL structures, describe what you need and let the AI write the query.

What the AI Assistant Can Do

• Generate Elasticsearch/OpenSearch queries: Search DSL, aggregations, filters
• Generate DynamoDB PartiQL: SELECT, INSERT, UPDATE, DELETE statements
• Explain query syntax: Ask "what does this query do?" for explanations
• Optimize queries: Get suggestions for query improvements
• Debug errors: Paste error messages for troubleshooting help

Supported Providers

DocKit supports two AI providers:

Provider	Models	Best For
OpenAI	GPT-4, GPT-4o, GPT-3.5	General-purpose, best quality
DeepSeek	DeepSeek Chat, DeepSeek Coder	Cost-effective, code-specialized

Configuring OpenAI

1. Open Settings → AI Assistant tab
2. Select OpenAI tab
3. Fill in:

Field	Description	Example
Model	OpenAI model name	gpt-4o, gpt-4, gpt-3.5-turbo
API Key	Your OpenAI API key	sk-proj-...
Prompt	Custom system prompt (optional)	See below
Proxy	HTTP proxy URL (optional)	http://127.0.0.1:7890

4. Click Save & Enable

Getting an OpenAI API Key

1. Go to platform.openai.com
2. Sign in or create an account
3. Navigate to API Keys → Create new secret key
4. Copy the key (starts with sk-)
5. Paste into DocKit

Note: OpenAI requires payment. GPT-4 costs ~$0.03 per 1K tokens. GPT-3.5-turbo is cheaper (~$0.002 per 1K tokens).

Configuring DeepSeek

1. Open Settings → AI Assistant tab
2. Select DeepSeek tab
3. Fill in:

Field	Description	Example
Model	DeepSeek model name	deepseek-chat, deepseek-coder
API Key	Your DeepSeek API key	From DeepSeek platform
Prompt	Custom system prompt (optional)	See below
Proxy	HTTP proxy URL (optional)	http://127.0.0.1:7890

4. Click Save & Enable

Getting a DeepSeek API Key

1. Go to platform.deepseek.com
2. Sign in or create an account
3. Navigate to API Keys
4. Create and copy your key
5. Paste into DocKit

Note: DeepSeek is more cost-effective than OpenAI. deepseek-coder is optimized for code generation tasks.

Custom Prompts

The Prompt field lets you customize how the AI responds. Default prompts are optimized for database queries, but you can tailor them.

Default Prompt Behavior

DocKit automatically injects:

• Index/Table name: Current index or table context
• Schema/Mapping: Field names and types
• Database type: Elasticsearch, OpenSearch, or DynamoDB

This context helps the AI generate accurate queries matching your actual data structure.

Custom Prompt Example

You are a database expert. Generate queries for Elasticsearch/DynamoDB.
- Use only fields from the provided schema
- Explain the query before showing code
- Suggest optimizations when relevant

Tip: Keep prompts concise. Long prompts consume tokens and slow responses.

Proxy Configuration

If you're behind a firewall or need to route requests through a proxy:

1. Enter proxy URL: http://127.0.0.1:7890
2. Format: http://host:port
3. Supports HTTP and HTTPS proxies

Common proxy tools:

• Clash: Default port 7890
• V2Ray: Various ports
• Corporate proxies: Check your network settings

Using the AI Assistant

Once configured, access the assistant from the editor:

1. Click the AI icon (✨) in the toolbar
2. Or press the keyboard shortcut for your platform
3. Type your question in natural language
4. Press Enter to send

Example Queries

Elasticsearch:

"Find all users who signed up in the last 7 days"
"Aggregate sales by region with average price"
"Search products containing 'laptop' sorted by price"

DynamoDB:

"Get all orders for user ID 'user-123'"
"Find items where status equals 'pending'"
"Update the email field for user 'alice'"

The AI responds with:

1. Generated query code
2. Brief explanation
3. Any suggestions or warnings

Applying Generated Queries

After the AI generates a query:

1. Review the code in the chat panel
2. Click Copy to Editor to insert into your query file
3. Edit as needed (adjust field names, add filters)
4. Execute with Cmd+Enter (or Ctrl+Enter)

Privacy and Data Handling

What DocKit sends to AI providers:

• ✅ Your text prompt (natural language question)
• ✅ Current index/table name
• ✅ Schema/mapping (field names and types)
• ❌ Your actual data (never sends database contents)
• ❌ Credentials (never sends API keys or passwords)

Data retention:

• OpenAI/DeepSeek may retain prompts for 30 days (check their policies)
• DocKit does not store your prompts locally after the session

Best practices:

• Don't paste sensitive data values in prompts
• Use generic examples when asking about production data
• Review generated queries before executing on production

Troubleshooting

Invalid API Key

Error: Invalid API key or Unauthorized

Solution:

1. Verify the key is correct (no extra spaces)
2. Check the key hasn't expired
3. Ensure sufficient credits/quota
4. Try regenerating the key

Connection Failed

Error: Network error or Timeout

Solution:

1. Check internet connection
2. Configure proxy if behind firewall
3. Verify API endpoint is accessible
4. Check if OpenAI/DeepSeek services are operational

Poor Quality Responses

Problem: AI generates incorrect or nonspecific queries

Solution:

1. Switch to a better model (GPT-4 instead of GPT-3.5)
2. Be more specific in your prompt (include field names)
3. Add schema context manually if auto-injection fails
4. Use deepseek-coder for code-specific tasks

Rate Limits

Error: Rate limit exceeded

Solution:

1. Wait a few seconds before retrying
2. Reduce prompt frequency
3. Upgrade your OpenAI/DeepSeek plan for higher limits

Switching Providers

To switch between OpenAI and DeepSeek:

1. Open Settings → AI Assistant
2. Configure both providers (they can be set up simultaneously)
3. Only one can be enabled at a time
4. Click Save & Enable on your preferred provider

The enabled provider handles all AI requests until you switch.

Disabling the AI Assistant

To turn off AI features:

1. Open Settings → AI Assistant
2. Click Reset on the active provider
3. Or clear the API key field

When disabled, the AI icon is grayed out in the toolbar.

Cost Management

AI queries consume tokens. Estimate costs:

Provider	Model	Cost per 1K tokens	Typical Query Cost
OpenAI	GPT-4o	$0.005	~$0.01-0.02
OpenAI	GPT-4	$0.03	~$0.05-0.10
OpenAI	GPT-3.5-turbo	$0.002	~$0.003-0.005
DeepSeek	deepseek-chat	$0.001	~$0.002
DeepSeek	deepseek-coder	$0.002	~$0.003

Tips to reduce costs:

• Use GPT-3.5-turbo or DeepSeek for routine queries
• Reserve GPT-4 for complex queries requiring deep reasoning
• Keep prompts concise (shorter prompts = fewer tokens)
• Cache common query patterns in your saved files

Next Steps

• Settings Guide — Configure other DocKit options
• Elasticsearch Cluster Management — Work with indices
• DynamoDB PartiQL Editor — Learn PartiQL syntax