Frequently Asked Questions

Simple answers to common questions about Helpmaton

Getting Started

How do I create my first workspace?

Workspace creation
After signing up, you'll be asked to create your first workspace. A workspace is a separate space for your agents, documents, and team members. Your first workspace is credited with $2 to try AI without adding your own keys.

Multiple workspaces let you keep projects or clients separate, each with its own budget and access.

What do I need to get started?

Account setup
All you need is a Helpmaton account. The free plan is free forever (though severely limited), and no credit card is required. Your first workspace includes $2 in free credits so you can try AI right away.

To cover AI usage, you can either bring your own key (via OpenRouter) or buy credits through Helpmaton.

Once you create a workspace, you can set up agents and upload documents right away.

For detailed setup instructions, see the Getting Started Guide.

Is Helpmaton right for my team size?

Built for individuals and teams
Helpmaton works for solo builders and growing teams. Start with one workspace and one agent, then add more as you scale.

Flexible plans let you move up when you need more workspaces, agents, or collaboration features.

How is this different from building my own AI agents?

Less infrastructure, more outcomes
Helpmaton gives you workspaces, memory, usage tracking, and integrations without building it yourself.

Faster time to value - You can launch agents in minutes instead of weeks of setup and maintenance.

How do I create my first agent?

Agent creation
Open your workspace, click "Create Agent", then give it a name and instructions. You can choose tools and memory options, too.

Ready right away - Your agent is available as soon as you create it, and you can connect other apps whenever you want.

What's the difference between a workspace and an agent?

Workspaces vs Agents
Think of a workspace as a folder for a project. It holds your agents, documents, team members, budget, and settings. You might keep one workspace for personal projects and another for work.

An agent is the AI assistant inside a workspace. Each agent has its own instructions, memory, and skills. You can have several agents in one workspace.

What is OpenRouter and why do I need it?

OpenRouter explained
OpenRouter gives you one key to access multiple AI providers (OpenAI, Anthropic, Google, and more).

Why it matters - If you bring your own API keys, Helpmaton connects through OpenRouter. You can also skip keys and buy credits through Helpmaton instead.

Agents

How does agent memory work?

How memory works
Agents remember important details from your conversations and use them to answer better over time.

Memory retention - Free plans keep 48 hours of detailed memory plus 30 days of summaries, while Pro plans keep 240 hours plus 120 days. Older memories are cleaned up automatically.

Smart search helps agents find the right past details when responding.

Learn more in the Agent Memory System documentation.

Can agents work together?

Agent delegation
Yes. Agents can pass tasks to other agents in the same workspace so work gets shared.

Multi-agent collaboration - You can set up teams of agents that handle different parts of a task and track who did what.

What tools do agents have access to?

Built-in tools
Agents have built-in tools like document search, memory search, web search, web page reading, email sending, notifications, and task handoffs.

External integrations - You can connect more tools and services, then enable them per agent.

How do I configure an agent's personality?

Custom instructions
When you create or edit an agent, add instructions that guide its tone and behavior.

Personality settings help you match the agent to a specific role or brand voice.

What's the difference between web search and content extraction?

Web search vs content extraction
Web search lets your agent look up current information and articles, like using a search engine.

Content extraction lets your agent read and summarize a specific page when you give it a URL.

Both help your agents use up-to-date information.

Can I customize how memory summaries are written?

Custom summarization prompts
Yes. You can set per-agent prompts for each time grain (daily, weekly, monthly, quarterly, yearly) so summaries match your workflow.

Why it matters - Different agents can retain and summarize context in different ways.

Can agents run on a schedule?

Agent schedules
Yes. You can set cron-based schedules so agents run automatically for reports, monitoring, and recurring tasks.

Use cases - Weekly summaries, daily monitoring, and proactive alerts.

What are Agent Skills?

Agent Skills explained
Agent Skills are optional instruction blocks that teach your agents how to use their tools effectively. They provide step-by-step guidance, examples, and edge-case handling for specific workflows.

How it works - Enable tools (MCP servers or built-in tools like document search, web search, email) for your agent, then pick from available skills that match those tools. Each skill is only available when all its required tools are enabled and connected.

What they cover - Skills cover analytics (PostHog, HubSpot), support (Zendesk, Intercom), project management (Linear, Notion, Todoist), e-commerce (Shopify), CRM (Salesforce), and more.

Where to enable - Agent detail page → External tools → Skills.

Learn more in the Agent Skills documentation.

Evaluation & Quality

What are Judge evals?

Evaluation made simple
Judge evals score conversations so you can measure quality, validate outcomes, and catch regressions after changes.

Why it helps - It provides quick feedback before you ship updates.

How does evaluation sampling work?

Sampling controls
You can evaluate a subset of traffic instead of everything, which keeps costs predictable while still giving you reliable insight.

Balance - More sampling equals deeper insight; less sampling equals lower cost.

Workspaces

What are workspace credits used for?

AI provider costs
Workspace credits are for AI provider usage (OpenRouter, OpenAI, Anthropic, Google, etc.), not for your Helpmaton subscription.

Flexible billing - Use your own OpenRouter key and pay providers directly, or buy credits through Helpmaton.

Can I use my own API keys?

Using your own API keys
Yes. You can use your own OpenRouter API key (the current supported option for bringing your own key).

Direct billing - When you add a key, requests use it and you pay OpenRouter directly. Spending limits and usage tracking still apply.

How do spending limits work?

Granular control
You can set spending limits for a workspace or a specific agent, for daily, monthly, or yearly periods.

Automatic protection - The system checks limits before each request, so you stay within budget.

Can I set limits for individual agents?

Agent-level caps
Yes. You can set separate limits for individual agents so high-traffic agents don’t impact the rest of your workspace.

Flexible time spans - Choose daily, monthly, or yearly limits per agent.

How many team members can I add?

Team collaboration
Free plans support limited team members, while Pro plans include unlimited managers.

Permission levels - Team members can have different access: View, Edit, and Owner. Pro plans support multiple managers with shared access to all workspaces.

What are external tool connections?

External tool connections explained
External tool connections let you connect agents to other apps and services you already use, like calendars, email, or databases.

Why it's useful - Agents can do real work, not just answer questions. They can schedule meetings, send emails, look up data, and more.

Billing & Pricing

What's included in my subscription?

Platform services
Your monthly subscription covers the Helpmaton platform: hosting, agent setup and management, memory, usage tracking, workspaces, team collaboration, document storage, and core features.

What's included - Everything you need to run and manage your AI agents on the Helpmaton platform.

Is there a free plan or trial?

Free plan access
The free plan includes 1 workspace and 1 agent and is free forever (though severely limited). No credit card is required.

Upgrade anytime - When you're ready, move to a paid plan to keep your agents running.

What if I need more than what's in my plan?

Upgrade anytime
You can move to a higher plan as soon as you need more workspaces, agents, or usage limits.

Flexible scaling - Start small, then expand when your workflows grow.

Are AI provider calls included in my subscription?

Separate costs
No. AI provider calls (OpenRouter, OpenAI, Anthropic, Google, etc.) are separate from your subscription.

Two options - Use your own OpenRouter key and pay providers directly, or buy credits through Helpmaton for pay-as-you-go billing.

How much do web searches cost?

Free daily allowance
All plans include 10 free web searches per 24 hours. Additional searches are billed at $0.008 each.

Transparent pricing - This keeps costs predictable for research and monitoring.

What counts as an AI request?

Understanding AI requests
An AI request is each time your agent processes a message. One message and one reply count as a single request.

What's included - This covers all agent work like answering questions, searching documents, recalling memory, and web searches.

What are credits and how do they work?

Credits explained
Credits are a prepaid balance for AI usage. When you buy credits, that balance goes down as your agents make AI requests.

How it works - Each request costs a small amount based on the work done. You can see costs in your usage reports, and credits are deducted automatically.

Should I use my own API keys or purchase credits?

Choosing the right option
Use your own API keys if you already pay AI providers directly and want to keep it that way.

Purchase credits if you want the simplest setup with one payment method.

Both options give you access to all Helpmaton features. Choose the billing style you prefer.

How does credit reservation work?

Automatic credit management
Before an AI request, the system estimates the cost and sets aside credits. After the request finishes, it adjusts to the real cost automatically.

Accurate billing - If the actual cost is lower, you get a refund. If it's higher, you pay the difference.

Learn more in the Credit System documentation.

What does 'subscription sharing' mean?

Subscription sharing explained
Subscription sharing (on Pro plans) lets multiple people manage the same Helpmaton account. All managers can access all workspaces and work together.

Why it's useful - It’s great for teams that want to share one Pro subscription.

What does 'working memory' and 'summaries' mean?

Memory types explained
Working memory is the detailed, recent memory your agent keeps. It remembers the last 48 hours (Free) or 240 hours (Pro).

Summaries are shorter notes from older conversations. They are kept for 30 days (Free) or 120 days (Pro).

This keeps agents helpful without storing everything forever.

API & Integrations

How do I access the API?

API for developers
Helpmaton offers a full API with documentation. You can sign in with web sessions, API tokens, workspace API keys, or secure sign-in (OAuth).

Complete documentation - See the API Reference for details and examples.

See also the API Reference documentation in the repository.

Can I set up webhooks?

App connections
Yes. Every agent can connect to other apps and services. You can send messages and receive AI-powered responses.

Real-time responses - Supports standard replies and live updates while the agent works.

Do you support streaming responses?

Streaming support
Yes. Helpmaton supports streaming responses (SSE) so you can build fast, real-time chat experiences.

Best for - Interactive UIs and tools that show answers as they are generated.

What are MCP servers?

External tool connections
MCP servers are connections to external tools, databases, and services. They let agents access information and take actions beyond built-in tools.

Configuration - Set up authentication, then enable the connection per agent. Your agents can use the tools you connect.

See the integration documentation for detailed setup instructions.

How do I connect my agents to Gmail, Google Calendar, or Google Drive?

Easy secure sign-in
Connect Google services in a few clicks. No complex setup—just connect your Google account.

Setup process - Go to "Integrations" in your workspace, create a connection for Gmail, Calendar, or Drive, click "Connect Google Account", sign in, and grant permissions. Enable the connection for your agents and you're done.

What agents can do - Gmail: search and read emails. Google Calendar: manage events and schedules. Google Drive: read files, Docs, Sheets, and Slides.

Secure by default - All connections use secure sign-in to keep your data private.

How do I connect my agents to Notion?

Easy secure sign-in
Connect Notion in a few clicks. No complex setup—just connect your Notion account.

Setup process - Go to "Integrations" in your workspace, create a connection for Notion, click "Connect Notion Account", sign in, and grant permissions. Enable the connection for your agents and you're done.

What agents can do - Read and search pages, create and update content, and work with databases. Great for knowledge and documentation workflows.

Secure by default - All connections use secure sign-in to keep your data private.

How do I deploy agents as Slack or Discord bots?

Chat platform integration
Go to the Integrations page in your workspace. For Slack, you can generate an app configuration. For Discord, you set up slash commands.

Secure by default - Both platforms use verification to ensure only valid requests are processed. Bots respond in real time and can use the agent’s tools and memory.

See the Slack Integration guide and Discord Integration guide for detailed setup instructions.

What does it mean to deploy an agent as a Slack or Discord bot?

Chat bot deployment explained
When you deploy an agent as a Slack or Discord bot, it joins your chat and people can talk to it there, like messaging a teammate.

What it enables - Your team can ask questions and get help without leaving Slack or Discord. The agent replies quickly, remembers context, and can use its tools right in chat.

It’s like having an assistant in your chat room.

Technical

How are errors handled?

Automatic error tracking
When something goes wrong, you’ll see it right away. Conversations are saved and errors are clearly marked with details.

Error visibility - Error badges make problems easy to spot, and detailed views help you fix issues faster.

What file formats are supported for documents?

Document management
Helpmaton supports markdown and text documents for knowledge bases. You can upload files and organize them in folders.

Smart search - Agents can search by meaning, not just keywords, to give better answers.

How does web search work?

Real-time information access
Agents can search the web for current information and summarize specific pages when you give a URL.

Easy enablement - You can turn these features on per agent with simple toggles.

What are conversation attachments?

Files and images in chat
Conversation attachments let you add files and images to a specific conversation so the agent can use them as context.

Different from documents - Knowledge base documents are Markdown/text; attachments are for per-chat files.

What infrastructure does Helpmaton use?

Reliable cloud infrastructure
Helpmaton runs on modern cloud infrastructure built for reliability and security. Data is encrypted and stored securely.

High availability - The platform scales with your usage so agents stay available when you need them.

Region - Deployed in London (eu-west-2) with built-in redundancy.

See the Architecture Overview for detailed technical information.

How does usage analytics work?

Comprehensive tracking
Helpmaton tracks your AI usage and aggregates it by day and hour. You can filter by date ranges to see history.

Detailed breakdowns - View usage and costs by workspace or agent to see where time and spend go.

Self-Hosting & Source-Available

Is Helpmaton open source or source-available?

Source-available license
Yes. Helpmaton uses the Business Source License 1.1, which makes the source code available for evaluation, internal use, and contribution.

License terms - It remains source-available until November 1st, 2029, when it converts to Apache License 2.0. You can view and contribute on GitHub.

See the LICENSE file for full license terms.

What license does Helpmaton use?

Business Source License 1.1
Helpmaton uses the Business Source License 1.1, which makes the software source-available while protecting commercial hosting rights until the Change Date.

Change Date - On November 1st, 2029, the license converts to Apache License 2.0. Until then, you can use it internally or self-host for your own business, but you cannot offer it as a hosted service to others.

Read the full license terms in the LICENSE.md file.

Can I self-host Helpmaton?

Full self-hosting support
Yes. The full source code is available on GitHub, so you can run your own instance and customize it.

Infrastructure requirements - You’ll need an AWS account with services like Lambda, DynamoDB, S3, API Gateway, CloudFront, and Route53. The platform uses Architect Framework for deployment. See the deployment documentation for setup steps.

Full control - Self-hosting gives you full control of your data and infrastructure.

What do I need to self-host Helpmaton?

Infrastructure setup
You’ll need an AWS account with services like Lambda, DynamoDB, S3, API Gateway, CloudFront, Route53, and ACM for SSL certificates. The platform is designed for AWS serverless infrastructure.

Development setup - Local development uses Architect Sandbox (DynamoDB emulator) and s3rver (local S3). See the development setup guide for details.

Deployment - Deploy via GitHub Actions or manually with Architect Framework. Required environment variables and secrets are listed in the deployment guide.

How does the vector database work?

How memory storage works
Agent memory is stored in specialized databases. Each agent organizes memory by time periods to make searching fast and relevant.

Reliable storage - Data is written safely and consistently to avoid loss or corruption.

Fast retrieval - Smart search helps agents quickly find relevant details from past conversations.

See the technical documentation for complete technical details.