The Alchemist

The Alchemist is your AI-powered builder assistant, available everywhere in the Chipp platform. It can analyze your chatbot’s conversations, configure app settings, generate content, manage knowledge sources, and more — all through natural language. Think of it as a co-pilot that knows every feature in the platform and can operate them for you.

Where to Find It

The Alchemist panel appears in two places:

• Dashboard: The panel on the main dashboard page operates in workspace mode. It has access to analytics across all your apps, can compare performance, and navigate you into any app’s builder.
• App Builder: When you’re editing a specific app, the Alchemist panel has full access to that app’s configuration. It can read and modify settings, knowledge sources, integrations, tags, and more.

What It Can Do

Conversation Analysis

The most powerful use case: ask the Alchemist to analyze your chat logs and tell you how to improve your app. It will:

1. Sample recent conversations across different categories (short/abandoned, long/engaged, recent, random)
2. Spin up a dedicated analysis agent with a fresh context window focused entirely on your conversation data
3. Identify frustration patterns, knowledge gaps, error patterns, and what’s working well
4. Return prioritized, actionable recommendations grounded in real conversation examples

Try asking:

• “Analyze my conversations and tell me how to improve this app”
• “What are users struggling with?”
• “Look at the last 7 days of chats and find knowledge gaps”

After the analysis, the Alchemist can immediately act on its own recommendations — updating your system prompt, adding knowledge sources, or configuring new features.

Analytics & Insights

Get instant answers about your chatbot’s performance without navigating dashboards:

What you can ask	What it does
”How many conversations did I have this week?”	Returns session counts, message volume, and trends
”What are my users talking about?”	Samples recent first messages and clusters them by topic
”Show me my top consumers”	Consumer report with engagement metrics, leads, and email collection
”Which app is getting the most traffic?”	Per-app overview with chat counts and consumer counts
”Search for conversations about refunds”	Full-text search across all chat messages
”What tags are being triggered most?”	Tag distribution with percentages and sample messages

All analytics tools accept a date range (e.g., “last 7 days”, “last 90 days”) and can be filtered to a specific app.

App Configuration

The Alchemist can read and modify every aspect of your app’s configuration:

• System prompt: “Update the system prompt to be more friendly and add instructions for handling billing questions”
• Model selection: “Switch this app to Claude Sonnet”
• Brand & styling: “Set the primary color to dark blue and generate a logo”
• Knowledge sources: “Add this URL as a knowledge source” or “Add these 5 URLs to the knowledge base”
• Custom actions: “Create an HTTP action that calls my CRM API to look up customer info”
• Tags: “Set up lead qualification tags for hot leads, cold leads, and support requests”
• Auth settings: “Require users to log in before chatting”
• Voice settings: “Enable voice mode with a professional female voice”
• Heartbeat: “Set up a daily check-in message that asks users how their project is going”
• Publishing: “Publish this app live”

Content Generation

Create visual assets and web content through conversation:

• Logos: “Generate a logo for a fitness coaching chatbot” — creates an AI logo and applies it
• Custom CSS themes: “Design a dark mode theme with purple accents and rounded bubbles” — generates and applies a full CSS theme via the Style Studio
• Landing pages: “Build a landing page that explains what this chatbot does” — creates a multi-page site via Chipp Sites
• Custom components: “Create a pricing table component with three tiers” — generates interactive HTML/CSS/JS components

Integration Management

Build and deploy integrations without writing code:

• MCP servers: “Create an MCP server that connects to my project management API”
• Deploy & test: “Deploy the integration and test the list_projects tool”
• OAuth setup: Navigate to OAuth configuration flows when needed

Marketplace & Templates

Discover and use pre-built app blueprints and composable skills:

• Browse: “Show me templates for customer support bots”
• Create from template: “Create a new app from the Sales Closer template”
• Apply skills: “Add the Empathetic Listener skill to this app” — composable prompt modules that enhance your app

Advanced Reasoning

For complex questions, the Alchemist can think more deeply:

• Deep thinking: Breaks down complex questions into structured reasoning steps. “Plan out how I should restructure my app to handle both sales and support.”
• Deep research: Spawns parallel web research agents to find authoritative information. “Research best practices for AI customer support bots in healthcare.”

Example Workflows

”Improve my app” workflow

This is the most common and most powerful workflow:

New app setup workflow

Cross-app analytics workflow

From the dashboard, you can analyze performance across all apps:

1. “Which of my apps is getting the most conversations?”
2. “Show me the chat analytics for the last 30 days across all apps”
3. “What are the most common topics users ask about?”
4. “Open the Customer Support Bot so I can make changes” — navigates you into the app builder with the Alchemist panel open

Long Conversations

When you work with the Alchemist for extended periods — building out a complex app, walking through analytics across many sessions, or chaining a long sequence of changes — the conversation history can grow large enough that the AI starts to lose focus on the most recent details. The Alchemist now handles this automatically with server-side context compaction.

What you’ll see in the chat

Once a conversation gets long enough that summarization becomes worthwhile, you’ll see:

1. A brief “Compacting earlier conversation…” indicator while the summary is being generated.
2. A summary divider that replaces the older messages — clearly labeled Earlier conversation summary — N messages compacted with the AI’s own summary of what was done before. You can expand it to read the summary; the divider stays in place for the rest of the session.
3. The chat continues normally below the divider, with the most recent ~7 user/assistant exchanges kept verbatim.

You don’t need to do anything to trigger compaction. It runs in the background, transparently, the moment the conversation needs it.

Recall: getting verbatim details back from the summary

A summary is, by design, a lossy compression of what happened earlier. If you ever ask the Alchemist about something specific that lives in the summarized portion — an exact file path, a value it generated, a tool result, the wording of an earlier instruction — it has a tool called recall_earlier_conversation that searches the full pre-compaction history (preserved as a searchable archive scoped to your current session).

You don’t need to invoke this tool manually. The Alchemist decides when to use it based on your question. For example:

• “What was that bot token we set up earlier?” → searches the archive, finds the verbatim mention
• “Earlier you generated a logo — what was the prompt?” → pulls back the exact tool call
• “Can you remind me what error we hit when we tried to deploy?” → returns the message snippet

If the archive doesn’t contain a match, the Alchemist will tell you so plainly rather than guess.

Which models support it

The native server-side compaction runs on Anthropic’s reasoning models — the Alchemist’s default (Claude Sonnet 4.6) is one of them, so most users get this behavior automatically. If your workspace has overridden the Alchemist model to a non-Anthropic option (GPT-5, Gemini, etc.), the Alchemist falls back to a similar built-in summarization step that produces the same divider and the same searchable archive — only the underlying mechanism differs.

Tips for long sessions

• Trust the divider. Once you see a compaction summary, the older messages still exist in the searchable archive. You haven’t lost them — they’re just no longer in the AI’s active working memory.
• Be specific when recalling. “What was the X we configured?” works better than “what did we do earlier?” because the recall tool uses keyword search.
• Start fresh if you switch contexts. If you finish one task and want to start something completely unrelated, click New chat in the Alchemist header. A fresh context window is faster than working through a long compacted history when the topics don’t relate.

Dashboard Mode vs App Mode

Capability	Dashboard	App Builder
Analytics (all apps)	Yes	Yes (filtered to current app)
Conversation analysis	Yes (any app)	Yes (current app)
Browse templates	Yes	Yes
Create app from template	Yes	Yes
Read app config	After selecting an app	Automatic
Modify app settings	After selecting an app	Automatic
Generate logos, CSS, sites	After selecting an app	Automatic
DOM interaction (uploads, OAuth)	After navigating to app	Yes

In dashboard mode, the Alchemist automatically calls select_app when it needs to work with a specific app. You can also tell it which app to work with: “Select my Customer Support Bot and update its system prompt.”

Tips

• Be specific: “Add a knowledge source about our refund policy” is better than “improve my knowledge base”
• Chain requests: The Alchemist maintains context, so you can say “Now also add one for the pricing page” after adding a knowledge source
• Use analysis first: Before making changes, ask the Alchemist to analyze conversations. Data-driven improvements beat guesswork.
• Filter by date: “Analyze conversations from the last 7 days” focuses on recent patterns. “Last 90 days” gives a broader picture.
• Focus your analysis: “Analyze conversations with a focus on onboarding issues” tells the analysis agent to prioritize a specific area.