How to Use Claude as a Personal Assistant
Claude works well as a personal assistant when you give it a clear work brief, recurring tasks, and review checkpoints. Here is how to set it up so the work builds on itself instead of starting from scratch each session.
Updated 5 min read
Claude can act as a personal assistant for tasks like drafting and editing, research and summarizing, scheduling decisions, and working through complex problems. The key to making that work well is giving Claude the right starting context each session, so it does not spend the first few exchanges re-learning your situation.
A tool like fenbs keeps your tasks, decisions, and where you left off in one record outside any single chat. Claude reads that record at the start of each session, so the context arrives before the work, not in the middle of it.
Write a standing work brief
A standing brief is the information Claude needs before any task: who you are, what kind of work you do, your current focus areas, and your preferences. It is not a long biography. It is the short paragraph you would give a capable new assistant on their first day.
Include the things that affect almost every task: the tone you prefer in written work, decisions you have already made that should not be re-opened, and anything Claude should always or never do. Keep it short enough to read in thirty seconds.
Define recurring tasks clearly
For tasks you do regularly, write a description once and reuse it. A weekly status update, a research brief, or a review checklist should describe the starting materials, the expected output, and any constraints. The more specific the description, the less Claude needs to ask.
- What is the starting point: "Here is the raw meeting transcript."
- What should the output be: "A bullet-point summary with action items, owner, and due date for each."
- What constraints apply: "Keep it under one page. Do not include anything that was marked as off the record."
- What happens after: "Save the summary as a task note under the meeting project."
Keep decisions outside the chat history
Chat history disappears when a session closes or a context window fills up. Any decision you made in a previous conversation, such as the format you agreed on for reports or the vendor you chose not to use, is gone unless you saved it somewhere.
In fenbs, Decisions and rules are shown to each connected assistant before it does anything else. That means a preference you recorded once is respected in every future session, not just the one where you mentioned it. Add a new rule when a decision matters enough to affect future work.
Give Claude the same starting point every session
fenbs keeps your tasks, Decisions and rules, and Where we left off in one record. Connect Claude over MCP from Integrations, and each session begins with the same context: what is in progress, what was decided, and where you stopped last time. Skills and prompts let you store recurring task templates any connected assistant can read.
Create a free accountFree for one person · No credit card · Works with Claude, ChatGPT, Cursor and more
Set review checkpoints for important work
A personal assistant should not publish, send, or submit anything without your review. Set that boundary explicitly in your instructions: "Draft the email and show it to me before sending." "Prepare the summary and ask me which sections to expand before writing the full version."
The review step is not an inefficiency; it is what makes delegation safe. Claude is accurate on many tasks but will occasionally misread a context or miss a nuance. A short review keeps the error small.
What Claude does and does not do as a personal assistant
Claude handles text work well: drafting, editing, summarizing, explaining, researching, and working through decisions. It can help you plan a project, prepare for a meeting, or learn something new. It does not have access to your email, calendar, or files unless you paste them in or connect a tool that provides access.
Claude in Claude.ai and in Claude Code are different surfaces with different capabilities. Claude Code is designed for software development tasks; what is Claude Code explains the distinction.
Claude.ai Projects provides a shared context within a project for subscribers on supported plans, which helps when you use Claude for an ongoing body of work. For team members who each use their own AI tools, a shared external record is more flexible than relying on one product's built-in memory, since it does not tie the shared information to a single subscription or model.
When Claude declines to answer something or asks for clarification, treat it as a useful signal. It may mean the task description was incomplete, the instruction was ambiguous, or the request falls outside Claude's capabilities for that surface. Adjust the instruction and try again rather than rephrasing in a way that obscures the original intent.
How to do this with fenbs
- Create a free account and add a project for your personal work.
- Write three to five Decisions and rules that capture your standing preferences: tone, format, things Claude should always ask about rather than decide.
- Add your current tasks with enough context that Claude can pick one up without a long explanation.
- Connect Claude over MCP from Integrations. Start each session by asking Claude to read the project context first.
- Before each session ends, ask Claude to update Where we left off with what was done and what comes next.
Building a routine also means knowing when to take a task back. If Claude consistently misses a nuance in a certain type of task, that task stays with you until you have written an instruction specific enough to prevent the miss. The routine is not just "hand everything to Claude"; it is a deliberate allocation of which work each session covers and where the review step sits.
For more on AI personal assistant options, see AI personal assistant. For information on Claude's general capabilities, see what is Claude AI.