What Makes an AI Assistant Useful
A language model on its own can answer questions and generate text, but it cannot interact with the world. It cannot visit a website, create a file, run a calculation, or generate an image. Tools are what bridge the gap between conversational AI and practical utility.
Moltbot comes with a rich set of built-in tools that give your AI assistant the ability to take real actions. Combined with the skills system that bundles tools into focused capability sets, Moltbot transforms from a chatbot into a genuine productivity assistant.
This guide walks through the major tool categories, explains how skills work, and covers how the Finna platform enhances these capabilities with managed infrastructure.
Browser Tools
Web Browsing
Moltbot's browser tools let your assistant visit websites, read content, interact with web applications, and extract information. This is not simple URL fetching - it is full browser automation powered by Playwright, meaning the assistant can handle JavaScript-rendered content, fill out forms, click buttons, and navigate multi-page workflows.
Common use cases include:
- Research: Ask your assistant to read an article, documentation page, or product listing and summarize the key points
- Data extraction: Pull pricing tables, contact information, or specifications from websites
- Form filling: Automate repetitive web form submissions
- Monitoring: Check a webpage for changes or specific content
Screenshot Capture
The assistant can take screenshots of web pages, which is useful for visual verification, documentation, or sharing what a page looks like without sending a link. Combined with the AI's visual understanding capabilities, screenshots become a two-way tool - the assistant can both capture and analyze visual content.
How Finna Enhances Browser Tools
Browser automation requires a running browser instance, which consumes significant memory and CPU. On a self-hosted setup, you need to install and configure Chrome or Chromium, manage memory limits, and handle browser crashes.
Finna provides a managed browser pool running on dedicated infrastructure. Each tenant gets an isolated Chrome instance with its own user data directory and CDP (Chrome DevTools Protocol) port. The browser pool is pre-configured and always available - your assistant's browser tools work out of the box without any setup on your part.
This separation also improves security. The browser runs on different infrastructure from your gateway, so even if a malicious website attempts to exploit the browser, it cannot access your gateway's file system or configuration.
File Management
Reading and Writing Files
Moltbot can create, read, edit, and organize files on its workspace volume. This is useful for:
- Document creation: Generate reports, summaries, meeting notes, or any text-based document
- Data processing: Read CSV or JSON files, process the data, and output results
- Configuration management: Edit configuration files based on natural language instructions
- Note taking: Store information for later retrieval across conversation sessions
File Organization
Ask your assistant to organize files by date, type, or project. It can rename files according to a naming convention, move files between directories, and create folder structures based on your preferences.
Workspace Isolation
On Finna, each gateway has a dedicated encrypted volume mounted at /data/workspace. This is your assistant's working directory - files created here persist across restarts and are accessible only to your gateway. The volume is encrypted at the infrastructure level, and the entire volume is deleted if you deprovision your gateway.
Web Search
Finding Information
Moltbot integrates with web search providers to find current information. Unlike the assistant's training data, which has a knowledge cutoff, web search provides access to real-time information - current news, recent publications, live pricing, and up-to-date documentation.
Search-Enhanced Responses
When your assistant encounters a question it cannot confidently answer from its training data, it can automatically search the web for current information. This combination of deep knowledge and real-time search produces responses that are both well-reasoned and factually current.
Research Workflows
For deeper research tasks, the assistant can perform multiple searches, cross-reference results, visit promising links for full content, and compile findings into a structured document. This turns a simple search query into a comprehensive research brief.
On Finna, web search tools come pre-configured with API keys for search providers like Brave Search, so you do not need to set up your own search API access.
Code Execution
Running Code
Moltbot can write and execute code in multiple languages. This is not limited to generating code snippets - the assistant can actually run the code and return results. Practical applications include:
- Calculations: Complex math, financial modeling, unit conversions
- Data analysis: Process datasets, generate statistics, create visualizations
- Prototyping: Write and test small programs or scripts
- Automation scripts: Create one-off scripts for file processing, data transformation, or API interaction
Safe Execution
Code execution happens within the gateway's isolated environment. On Finna, this means code runs inside your dedicated Firecracker microVM with its own file system and resource limits. A runaway script cannot affect other tenants or the host infrastructure.
Language Support
The gateway environment includes common runtime tools, but the primary execution environment is designed for JavaScript/TypeScript given Moltbot's Node.js foundation. For other languages, the assistant can use its browser tools to access online code execution environments.
Image Generation
Creating Visual Content
Moltbot can generate images through integrated AI image generation providers. Describe what you want - a logo concept, a diagram, an illustration, a social media graphic - and the assistant creates it.
Practical Applications
- Social media content: Generate post images, story graphics, or profile pictures
- Presentations: Create custom illustrations for slides
- Mockups: Quick visual mockups for design discussions
- Creative projects: Artwork, concept art, visual brainstorming
Model Flexibility
Since Moltbot supports multiple AI providers, you can configure which image generation model to use based on your quality requirements and budget. Switch between providers without changing your workflow.
Text-to-Speech
Voice Generation
Moltbot integrates with text-to-speech services to generate audio from text. This is useful for:
- Voice messages: Generate spoken responses in messaging channels that support audio
- Content creation: Create narration for videos, podcasts, or presentations
- Accessibility: Convert text content to audio for easier consumption
- Language learning: Hear correct pronunciation of words and phrases
Provider Integration
On Finna, text-to-speech is pre-configured with API access to providers like ElevenLabs and OpenAI TTS. The assistant can generate natural-sounding speech in multiple languages and voice styles without requiring you to set up API keys or manage provider relationships.
The Skills System
What Are Skills
Skills in Moltbot are curated bundles of tools and instructions designed for specific use cases. Rather than enabling individual tools one by one, a skill activates a coherent set of capabilities along with behavioral guidelines for using them effectively.
Think of skills as roles your assistant can take on. A "Research Assistant" skill might bundle web search, browser tools, and file management with instructions to always cite sources and organize findings by topic. A "Content Creator" skill might combine image generation, text-to-speech, and file management with guidelines for brand consistency.
Built-In vs Custom Skills
Moltbot ships with a set of built-in skills covering common use cases. On the Finna dashboard, you can enable or disable these skills with a toggle and see what tools and instructions each skill includes.
You can also create custom skills tailored to your specific workflow. Define which tools to include, write behavioral instructions, and save the skill for reuse. Custom skills are stored in your gateway's configuration and persist across restarts.
Skill Composition
Skills can be combined. Enable multiple skills simultaneously, and your assistant has access to all their tools and follows all their guidelines. This composability means you do not need a single monolithic configuration - build focused skills and combine them as needed.
Pre-Configured Tools on Finna
One of the main advantages of using Finna's managed platform is that many tools come pre-configured and ready to use:
- Browser pool: Managed Chrome instances with isolated user profiles
- Web search: Brave Search API access included
- Text-to-speech: ElevenLabs and OpenAI TTS configured
- Image generation: Available through configured model providers
These are platform-provided capabilities that work across all tenants without individual setup. You bring your own language model API key (Anthropic, OpenAI), and the platform provides the supporting tool infrastructure.
Tool Permissions and Safety
Controlling Access
Not every deployment needs every tool. A customer support assistant probably does not need code execution. A research assistant might not need file write access. Moltbot lets you control tool availability at the configuration level - disable what you do not need to reduce your attack surface.
Confirmation Prompts
For potentially destructive actions (like deleting files or executing code), the assistant can be configured to ask for confirmation before proceeding. This adds a human-in-the-loop checkpoint for actions that cannot be easily undone.
Audit Trail
On Finna, tool usage is part of the audit log. You can review what tools your assistant used, when, and in what context. This visibility is valuable for both debugging and compliance.
Getting Started with Tools
The fastest way to explore Moltbot's tools is to deploy a gateway on Finna and start experimenting through your preferred messaging channel. Ask your assistant what it can do, try different tasks, and explore the skills configuration in the dashboard.
Start with the tools that map to your immediate needs - web search and browsing for research, file management for document work, or browser automation for web tasks. As you build familiarity, layer in additional capabilities through the skills system.
The goal is an assistant that fits your workflow rather than forcing you to adapt to its limitations. With the right combination of tools and skills, Moltbot becomes a genuinely useful member of your team.