Use cases

See how teams use Synjar to give their AI assistants access to knowledge.

Which mode should I use?

Synjar has two modes. Choose based on your use case:

Your scenario	Recommended mode	Why
100+ FAQ articles customers search	Search Links	AI needs to search, not read everything
5-10 brand voice guidelines	Instruction Sets	AI needs full context to write consistently
Research papers (hundreds of PDFs)	Search Links	Too large for full context, search on demand
Sales playbook (20 docs max)	Instruction Sets	Reps want complete, curated information
Support documentation	Search Links	Dynamic search for varied questions
Employee onboarding procedures	Instruction Sets	Step-by-step guides need full context
Growing policy library	Search Links	Scales with unlimited documents
Compliance documentation	Instruction Sets	Must follow procedures exactly

Most teams use both together

• Search Links for reference material (large knowledge bases)
• Instruction Sets for core instructions (guidelines AI must follow)

---

AI Customer Support

Recommended mode: Search Links for large FAQ databases, Instruction Sets for core policies

Use Search Links when:

• You have 50+ help articles
• Customers ask varied questions
• AI needs to search for specific answers

Use Instruction Sets when:

• You have 5-10 core policies (returns, shipping, warranties)
• Support needs consistent messaging on key topics
• AI should know all policies upfront

What to upload

Search Links (searchable reference):

• Product documentation
• FAQ pages
• Troubleshooting guides

Instruction Sets (always-loaded context):

• Return policy
• Shipping policy
• Core pricing rules

Example flow

Customer: "Can I return this after 30 days?"
                    ↓
Support Bot checks Instruction Set (return policy loaded)
                    ↓
Bot: "Our standard return window is 30 days. However,
      for defective items, we accept returns within 90 days..."

---

Internal Knowledge Search

Recommended mode: Instruction Sets for procedures, Search Links for reference

Use Search Links when:

• You have 100+ internal documents
• Employees need to find specific information
• Documents change frequently

Use Instruction Sets when:

• You have core procedures everyone must follow
• Information must be presented in specific order
• AI should always have access to key policies

What to upload

Instruction Sets (core procedures):

• Employee handbook highlights
• Key HR policies
• Security procedures

Search Links (reference library):

• Full policy documents
• Meeting notes and decisions
• Department-specific guides

Example queries

• "What's the PTO policy for part-time employees?"
• "How do I submit an expense report?"
• "What's the approval process for new vendors?"

---

Brand Guidelines & Marketing

Recommended mode: Instruction Sets (AI needs full context for consistent voice)

Why Instruction Sets?

• Brand voice requires complete context, not snippets
• Order matters: tone → vocabulary → examples → do's and don'ts
• AI should always have these guidelines loaded

What to upload

Instruction Sets (curated, ordered):

1. Brand voice overview
2. Tone guidelines
3. Vocabulary and terminology
4. Writing examples
5. Common mistakes to avoid

Integration

Add to Claude custom instructions:

Always follow our brand guidelines at: https://api.synjar.com/s/is_your_set

---

Sales Enablement

Recommended mode: Instruction Sets for playbooks, Search Links for reference

Use Instruction Sets when:

• Creating a complete sales playbook (5-20 docs)
• Reps need ordered, curated information
• Content includes objection handling, pricing, competitive info

Use Search Links when:

• You have 100+ case studies to search
• Need to find specific customer examples
• Product catalog is large

What to upload

Instruction Sets (sales playbook):

1. Product overview
2. Pricing guide
3. Competitive comparisons
4. Objection handling
5. Case study highlights

Search Links (reference):

• Full case studies
• Technical specifications
• Integration documentation

Example use

Sales Rep: "What differentiates us from Competitor X?"
                    ↓
AI checks Instruction Set (competitive comparisons loaded)
                    ↓
AI: "Three key differentiators: 1) Our semantic search
     is 40% more accurate, 2) We offer self-hosted..."

---

Documentation QA

Recommended mode: Search Links (large doc sets need dynamic search)

Why Search Links?

• Technical docs are typically 50-500+ pages
• Developers search for specific endpoints/examples
• Content updates frequently

What to upload

Search Links (all searchable):

• API documentation
• SDK guides
• Architecture docs
• README files
• Integration examples

Integration with Claude MCP

{
  "mcpServers": {
    "company-docs": {
      "command": "npx",
      "args": ["-y", "@synjar/mcp-server"],
      "env": {
        "SYNJAR_SEARCH_LINK": "https://api.synjar.com/s/sl_your_link_id"
      }
    }
  }
}

---

Research & Analysis

Recommended mode: Search Links (too many documents for full context)

Why Search Links?

• Research often involves hundreds of papers
• Need to find specific methods, findings, citations
• Can't load everything into context

What to upload

Search Links:

• Research papers (PDF)
• Reports and whitepapers
• Data documentation
• Literature reviews

Example queries

• "What methods were used to measure X in the 2023 studies?"
• "Which papers discuss the relationship between A and B?"
• "What are the main findings about Y?"

---

Legal & Compliance

Recommended mode: Instruction Sets for procedures, Search Links for contracts

Use Instruction Sets when:

• You have core compliance procedures (must follow exactly)
• Audit requirements need consistent responses
• Regulatory guidance must be followed precisely

Use Search Links when:

• Searching across many contracts
• Finding specific clauses
• Looking up regulatory references

What to upload

Instruction Sets (compliance procedures):

• Regulatory requirements summary
• Audit response procedures
• Data handling policies

Search Links (contract library):

• Contracts and agreements
• Full regulatory documents
• Legal precedents

Privacy

For sensitive documents, use self-hosted deployment or workspace isolation.

---

E-commerce Product Support

Recommended mode: Search Links (large catalogs need search)

Why Search Links?

• Product catalogs are large (100+ SKUs)
• Customers ask specific product questions
• Inventory and specs change frequently

What to upload

Search Links:

• Product specifications
• User manuals
• Compatibility guides
• Installation instructions

Example queries

• "Is this laptop compatible with USB-C monitors?"
• "What's the battery life of Model X?"
• "Can I use this printer with Mac?"

---

Integration Examples

Claude Desktop (MCP) - Search Links

For dynamic search across documents:

{
  "mcpServers": {
    "synjar-search": {
      "command": "npx",
      "args": ["-y", "@synjar/mcp-server"],
      "env": {
        "SYNJAR_API_KEY": "your-api-key",
        "SYNJAR_WORKSPACE_ID": "ws_your_workspace"
      }
    }
  }
}

Claude Desktop (MCP) - Instruction Sets

For always-loaded guidelines:

{
  "mcpServers": {
    "brand-guidelines": {
      "command": "npx",
      "args": ["-y", "@synjar/mcp-server"],
      "env": {
        "SYNJAR_INSTRUCTION_SET": "https://api.synjar.com/s/is_xyz789"
      }
    }
  }
}

ChatGPT (Custom GPT)

Add to your GPT instructions:

For product questions, search: https://api.synjar.com/s/sl_abc123?q={query}
For brand guidelines, always reference: https://api.synjar.com/s/is_xyz789

LangChain

from langchain.retrievers import SynjarRetriever

# For Search Links (RAG)
retriever = SynjarRetriever(
    api_key="your-api-key",
    workspace_id="ws_your_workspace"
)
docs = retriever.get_relevant_documents("What is the refund policy?")

# For Instruction Sets (full context)
import requests
instructions = requests.get("https://api.synjar.com/s/is_xyz789").text

Custom Integration

// Search Links - dynamic search
async function searchDocs(query) {
  const response = await fetch(
    'https://api.synjar.com/s/sl_your_link?q=' + encodeURIComponent(query)
  );
  return response.text();
}

// Instruction Sets - full context
async function getInstructions() {
  const response = await fetch('https://api.synjar.com/s/is_your_set');
  return response.text();
}

// Use together
const instructions = await getInstructions();
const context = await searchDocs(userQuestion);

const answer = await llm.complete({
  system: instructions,  // Always-loaded guidelines
  prompt: `Context: ${context}\n\nQuestion: ${userQuestion}`
});

---

Getting Started

1. Sign up for Synjar Cloud or self-host
2. Upload your documents
3. Decide: Search Links, Instruction Sets, or both?
4. Create Search Links for searchable content
5. Create Instruction Sets for curated guidelines
6. Connect to your AI application

Need help choosing? Contact us or check the FAQ.