A custom prompt turns Claude into a personal productivity assistant

Automating the morning brief

A user of the productivity app Todoist shared a method for using Claude to manage a complex workload. The user struggles with remembering to check the app regularly, which leads to an overwhelming number of overdue tasks. To solve this, they created a system where Claude acts as a personal assistant to summarize the day.

Every weekday morning, Claude generates a brief that covers what the day holds before the user even starts work. This includes a list of overdue items, tasks due today, and a lookahead at the next three days. The goal is to provide a calm, opinionated plan rather than just a raw dump of every single item on a list.

As How-To Geek reported, this setup helps the user manage multiple projects including writing, a small business, and social media channels. By having the AI do the heavy lifting of sorting through the data, the user can focus on execution rather than organization. The system is designed to be readable in under a minute to keep the morning routine efficient.

How the system connects data

The setup works by giving Claude read-only access to several different data sources. It can see the user's Todoist tasks, their calendar for fixed appointments, and their email accounts. It also has access to local folders where writing files are stored in 'In Progress' or 'Published' directories.

Claude uses this information to verify what work has actually been completed. For example, if a task involves writing an article and a matching file exists in the 'Published' folder, Claude flags it as likely finished. It does not delete or change anything automatically; it simply presents the evidence and asks for permission to mark the task as done.

The AI also looks for missing information that might cause problems later. It identifies commitments that lack a specific next step or tasks that have a real deadline but no assigned date. This proactive scanning ensures that nothing 'creeps up' on the user unexpectedly during the work week.

Triage and follow-up logic

One of the most useful features of this prompt is its ability to suggest new tasks based on current interactions. If the user replies to a partner about a project, Claude can notice the lack of a follow-up task and suggest creating one for a week later. The user then decides whether to accept or ignore these suggestions.

The prompt also handles 'Inbox triage' for new items added to the Todoist list. Instead of the user having to decide where a new task belongs, Claude suggests a project, a date, or a specific next step for every item in the inbox. This helps keep the main task list from becoming cluttered with unorganized thoughts.

The output is structured to prioritize the most important actions. It includes a '3x5 card' section which mimics a physical desk accessory the user designed. This section lists one 'must-do' item, two to four 'should-dos,' and a few 'easy wins' that fit around the user's existing calendar appointments.

Portability and customization

While the user currently uses Claude, they noted that the prompt is highly portable across different AI models. They have tested variations of this setup with other bots like Codex, Grok, and Hermes. As long as the chosen AI has the necessary connectors to access the apps, the logic remains the same.

The prompt is designed to be 'opinionated,' meaning it is instructed to make choices about what matters most. It tells the AI to respect dates the user has deliberately pushed out and to ignore items just because they are marked as high priority. This prevents the daily brief from being cluttered with non-urgent information.

The user mentions that they can have a conversation with the AI to refine the brief if they disagree with its logic. This interactive element allows the assistant to learn the user's preferences over time. It functions much like a human assistant who learns how the boss thinks and prefers to organize their day.

What this means for users

For people who feel overwhelmed by digital to-do lists, this setup offers a way to turn a passive list into an active plan. It moves the burden of organization from the human to the AI. By automating the review of emails and files, it reduces the mental energy required to stay on top of a busy schedule.

This approach is particularly useful for those who manage multiple roles or projects simultaneously. It ensures that small tasks don't fall through the cracks while keeping the focus on the most important goals. It also provides a bridge for those who still prefer a physical element to their workflow, like writing tasks on paper cards.

The next step for users interested in this is to identify which apps they use most frequently. They can then look for ways to connect those apps to an AI with read-only access. By creating a structured prompt that defines the 'rules' of the brief, anyone can begin to automate their daily planning.

Technical requirements

To replicate this, a user needs an AI model that can access external data. This usually requires using specific connectors or plugins that allow the AI to read from Todoist and email providers. The user specifically mentioned using read-only access to ensure security, as they do not trust the AI to send or delete emails autonomously.

The prompt itself is the core of the system. It provides specific instructions on how to handle different types of data, such as distinguishing between 'In Progress' and 'Published' folders. It also sets the tone for the interaction, requesting a 'calm' and 'opinionated' style to avoid a generic list of tasks.

Because the prompt is detailed, it covers many edge cases, such as how to handle tasks with no dates. By defining these rules clearly, the user ensures the AI behaves consistently every morning. This level of detail is what transforms a simple chat into a functional productivity tool that mimics a real-world assistant.

Next steps for the user

The user continues to refine the prompt as their workflow evolves. They have noted that the setup has changed significantly over several months of use. This suggests that the prompt is not a static piece of code but a living set of instructions that can be updated as new projects or habits emerge.

For those looking to try this, the source provides the full text of the prompt. It includes specific instructions for the AI to read the Todoist Inbox, check for work already finished, and suggest follow-ups. It also outlines the exact order in which the morning brief should be presented to the user.

The ultimate goal of this system is to reduce the 'me problem' of forgetting to check the app. By having Claude proactively reach out with a summary, the user stays on track without having to manually audit their own progress. It turns the AI into a proactive partner in managing a complex, multi-faceted life.

Implementation details

The prompt specifically instructs the AI not to change anything in Todoist on its own. It can only complete, create, move, or reschedule tasks after the user gives explicit approval in the conversation. This safety measure ensures that the user remains in control of their data and schedule at all times.

The AI is also told to look for 'watch-outs,' which are tasks that might sneak up on the user. These include commitments with no next step or items due soon that need preparation to start now. This proactive identification is a key difference between a simple list and a managed plan.

By asking the AI to provide evidence for finished tasks, such as a link to a published file or a sent email, the user can quickly verify the AI's logic. This creates a high-trust environment where the user can quickly say 'yes' to the AI's suggestions. It is a practical way to use AI to handle the 'busy work' of organization.

The 3x5 card strategy

A unique part of this workflow is the transition from digital to physical. The user uses a wooden tray and 3x5 cards to keep their most important tasks visible on their desk. Claude is tasked with selecting the specific items that should go on these cards each morning.

This physical list always starts as a digital list, but the AI helps filter it down to the essentials. By identifying one 'must-do' and a few 'should-dos,' the AI prevents the user from feeling overwhelmed by a long list of possibilities. It ensures that the most important work is what gets done first.

This hybrid approach combines the power of AI data processing with the psychological benefits of physical task management. It allows the user to benefit from the AI's ability to scan hundreds of tasks while maintaining a simple, focused workspace. It is a highly personalized way to manage a complex professional and personal life.

Summary of the prompt

The prompt provided by the user is comprehensive and covers every aspect of the morning brief. It defines the persona of the assistant, the data sources available, and the specific logic for identifying completed work. It even dictates the order of the sections to ensure a consistent experience every morning.

It explicitly tells the AI to ignore certain types of tasks, such as those the user has deliberately pushed out. This ensures the brief remains relevant and doesn't include noise. The prompt also includes a specific instruction to keep the entire brief readable in under a minute.

For anyone looking to copy this setup, the prompt serves as a complete blueprint. It shows how to move beyond simple questions and toward a system that understands context. By giving the AI clear rules for 'how to judge what matters,' the user has created a tool that truly understands their unique workflow.

Future possibilities

While the current setup is read-only for many apps, the user mentions they might eventually give Claude write access to complete tasks automatically. This would further reduce the friction of managing a to-do list, as the AI would handle the administrative side of checking off items without asking first.

The system could also be expanded to include more data sources, such as health data or more complex project management tools. As AI models become more capable of handling multiple tool connections, these types of personalized briefs will likely become more common and easier to set up.

The core lesson here is the power of specific prompting. By defining the 'shape' of the day and the rules for triage, any user can turn a general AI into a specialized assistant. It is a prime example of how to use current technology to solve common human problems like ADHD-related overwhelm and information overload.

Key takeaways

The main takeaway is that AI can be used to synthesize large amounts of personal data into actionable summaries. Instead of looking at a list of many items, the user looks at a brief of a few items. This shift from 'data' to 'information' is what makes the system effective for a busy person.

The system also highlights the importance of 'opinionated' AI. By telling the AI to make choices about what is important, the user avoids the trap of the AI just repeating everything it sees. The AI acts as a filter, which is the most valuable service a personal assistant can provide.

Finally, the setup demonstrates how to maintain human-in-the-loop control. By requiring approval for changes and providing evidence for completed tasks, the user ensures the AI remains a helpful tool rather than a source of errors. This balance of automation and oversight is the key to successful AI integration in daily life.

The role of context

Context is what makes this prompt work. Because Claude can see the user's writing folders, it knows that a task is done not just because the user said so, but because the file actually exists in the 'Published' folder. This level of verification is much more reliable than a simple manual check.

The AI also understands the context of the user's calendar. It knows when the user is busy and can suggest tasks that fit into the remaining open time. This means the '3x5 card' isn't just a list of things to do; it is a realistic plan that accounts for the user's actual availability.

By combining Todoist, calendars, emails, and local files, the AI creates a view of the user's life. This holistic view allows it to spot connections that a human might miss, such as a needed follow-up from a specific email thread. It turns the AI into a proactive coordinator of the user's time and energy.

Managing ADHD and overwhelm

The user explicitly mentions that their struggle with Todoist is a 'me problem' related to ADHD. They find it difficult to remember to check the app, which causes the list to grow stale and become overwhelming. This system is designed specifically to solve that problem by bringing the information to the user.

By having Claude prepare the brief every morning, the user doesn't have to go looking for their tasks. The information is waiting for them at their desk. This reduces the cognitive load of starting the day, as the user doesn't have to decide where to begin; the AI has already made those suggestions.

The 'watch-outs' section is particularly helpful for managing ADHD. It highlights things that are easy to forget, like tasks with deadlines but no dates. By surfacing these hidden responsibilities, the AI helps the user stay ahead of potential issues before they become urgent problems. It acts as an external brain for staying organized.

The power of read-only access

A key security and functional choice in this setup is the use of read-only access for emails and calendars. The user explains that they do not trust AI enough to give it write access to their email accounts. This allows the AI to gather information without the risk of it sending or deleting important messages.

Read-only access is sufficient for the AI to perform its primary job: information synthesis. It can see the content of emails to identify tasks, but it doesn't need to interact with the recipients. This creates a safe boundary where the AI can be helpful without being a liability.

This approach can be replicated by anyone who wants to use AI for personal organization. By keeping the AI in a 'read-only' role, users can enjoy the benefits of automated summaries while maintaining full control over their communications. It is a balanced way to integrate AI into a sensitive personal workflow.

Creating a custom brief

Anyone can create a similar brief by following the structure of this prompt. The key is to define the 'What,' the 'How,' and the 'Order.' The 'What' is the data (Todoist, email, etc.), the 'How' is the logic (check for published files, suggest follow-ups), and the 'Order' is the final format (the 3x5 card, the watch-outs).

The user's prompt is a great example of how to give an AI specific instructions to avoid generic outputs. By telling it to be 'calm and opinionated,' the user ensures the tone of the brief is appropriate for a morning routine. They also provide clear rules on what to ignore, which is essential for keeping the brief concise.

The prompt is also designed to be modular. A user could easily remove the email section or add a section for fitness goals. Because the instructions are so specific, the AI is less likely to hallucinate or provide irrelevant information. It turns the AI into a highly specialized tool tailored to the user's unique needs.

Summary of the workflow

The entire workflow can be summarized as: data collection, automated triage, and human-approved execution. Claude collects the data from various apps, triages it based on the user's rules, and presents a summary. The user then reviews the summary and gives the final approval to move tasks or mark them as done.

This cycle ensures that the AI does the heavy lifting of sorting through the noise, while the human remains the ultimate decision-maker. It is a perfect example of human-AI collaboration. The AI provides the 'what' and the 'when,' while the human provides the 'why' and the 'how.'

By following this model, anyone can create a personalized morning brief that makes them smarter and more productive every day. It moves the user from a state of reacting to their to-do list to a state of proactively executing a plan. It is a practical, scalable way to manage a complex life using current AI technology.

Final thoughts on AI assistants

This story highlights a shift in how people are using AI. Instead of just asking the AI to write poems or code, users are using it to manage the logistics of their lives. It is becoming a tool for cognitive offloading, where the AI handles the organization so the human can focus on the work.