Getting Started with Ditto: Build Your AI Memory in 5 Minutes

You’ve tried ChatGPT, Claude, Gemini. They’re powerful — until you close the tab and start over. Ditto is different: every conversation builds a persistent memory that makes the next one better.

This guide walks you through your first five minutes with Ditto — from first message to a fully contextualized AI workspace.

Step 1: Open Ditto

Head to assistant.heyditto.ai and sign in with your Google account. No credit card required — the free tier gives you enough tokens to explore every feature.

Ditto runs as a PWA (progressive web app), so it works in any browser. On mobile, tap “Add to Home Screen” for a native app experience on iOS or Android.

Step 2: Send Your First Message

Pick any model from the model selector — Google Gemini, Claude, GPT, or others — and start chatting. Ask about a project you’re working on. Explain something you’re interested in. Be specific:

I'm building a SaaS app for freelancers to track time and invoices.
The stack is Next.js, Supabase, and Stripe. I'm stuck on
webhook reliability — events sometimes arrive out of order.

Here’s what happens behind the scenes: Ditto saves the conversation as a memory and extracts subjects — “Next.js,” “Supabase,” “Stripe,” “webhooks,” “SaaS” — into your knowledge graph. This happens automatically. You don’t configure anything.

Step 3: Come Back Tomorrow and See the Difference

This is where Ditto clicks. Start a new conversation:

What were we discussing about webhook reliability?

Ditto searches your memories using learned retrieval weights — a composite of semantic similarity, recency, and frequency — and pulls in the relevant context. No re-explaining. No “I don’t have access to previous conversations.” It just knows.

You’ll see the retrieved memories displayed as expandable cards in the chat, so you can verify exactly what context the AI is working with. No black box.

Step 4: Create Your First Thread

Threads are Ditto’s core organizational unit — living workspaces that never go stale.

Click the + button in the sidebar to create a new thread. Name it something meaningful — “SaaS Project,” “Job Search,” “ML Research” — and optionally pick an icon.

Now attach context to your thread:

• Subjects: Pull in topics from your knowledge graph. Attach “Next.js” and “Stripe” to your SaaS thread, and every conversation in that thread will have deep context about those subjects before you type a word.
• Memories: Pin specific conversations that matter. Found the breakthrough answer about webhook ordering? Pin it to the thread so it’s always in context.
• Notes: Add freeform text the AI should know. “Launch deadline is April 15.” “Use Supabase Edge Functions, not serverless.” “Budget is $200/month for infrastructure.”

These attachments are injected into the AI’s system prompt every time you chat in the thread. They persist across sessions — come back in a week or a month and your thread is exactly as rich as you left it.

Step 5: Explore Your Knowledge Graph

After a few conversations, click Subjects in the sidebar to see your knowledge graph taking shape. Ditto automatically extracts people, topics, technologies, and concepts from your conversations and maps the connections between them.

The graph is interactive — click any subject to see every memory related to it, discover connections you hadn’t noticed, and understand how your conversations relate to each other.

This isn’t a flat list of facts like ChatGPT’s memory. It’s a structured graph that gets richer with every conversation. The more you use Ditto, the smarter it gets — across all your threads.

Step 6: Use AI Agents

Ditto comes with built-in agents that extend beyond text conversation:

• Research Agent: Ask Ditto to research a topic and it’ll search the web, read pages, and synthesize findings — all within your thread, all saved to memory.
• Art Agent: Generate images from descriptions. The results are stored in your conversation history alongside the text context.
• Sub-Agents: For complex multi-step tasks, Ditto can spawn sub-agents that work concurrently. You see their progress in real-time — no waiting for a black-box result.

Try it: ask Ditto to research something within a thread. The findings become part of your knowledge graph automatically.

Step 7: Connect Your Other AI Tools via MCP

This is the power move. Ditto implements the Model Context Protocol (MCP) as both a server and a client.

Ditto as MCP Server — Use your Ditto memory inside Claude, Cursor, Claude Code, or any MCP-compatible tool:

{
  "mcpServers": {
    "ditto": {
      "url": "https://api.heyditto.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Generate an API key in Settings > MCP API Keys. Now Claude can search your memories, read your personality profile, and save new memories back to Ditto. One knowledge graph powering all your AI tools.

Ditto as MCP Client — Connect external MCP servers to give Ditto new capabilities. Add servers in Settings, and their tools become available in your conversations.

For the full setup guide, see Connect Ditto to Any AI Assistant.

Step 8: Pick Your Plan

Ditto’s free tier lets you explore everything — memory, threads, knowledge graph, agents. When you’re ready for more, upgrade:

All plans include full access to every feature. The only difference is token volume — you’re never locked out of functionality.

What Happens Next

Here’s the part that’s hard to explain but easy to experience: Ditto gets better the more you use it.

After a week, your knowledge graph has dozens of subjects. Your threads carry context that would take pages to re-explain. Your memories form a searchable archive of every important conversation you’ve had.

After a month, switching to an AI that doesn’t remember you feels broken.

That compounding effect — more usage, more context, more value — is what makes Ditto different from every AI assistant that treats each conversation as disposable.

Ready to start building your AI memory? Open Ditto — it takes 30 seconds to sign in and your first conversation starts building context immediately.