Some meetings fail before they start. No agenda. No context. No clear decision to make. Everyone shows up, catches up for ten minutes, and leaves with a vague list of things to figure out later. Here's how to automate the preparation so the meeting does the work it's supposed to do.

This issue covers a five-tool AI workflow that runs twice a day in the background, scans your emails, texts, and calendar, updates your task list automatically, builds a structured agenda for every upcoming meeting, and generates a branded presentation from that agenda without you touching it. The tools: Day.ai as the data layer, Todoist for task management, Claude for the automation logic, Prompt Cowboy to build the prompts that run the system, and Gamma to turn the agenda into a visual deck. The result: every meeting you walk into has a BLUF, historical context, prioritized action items, and a presentation ready to go.

A project manager I know spent three hours before a client progress meeting pulling together what she needed: reviewing email threads to remember what had been promised, checking the task list to see what was overdue, building a slide deck that looked presentable, and writing an agenda that nobody had sent yet.

The meeting itself took 45 minutes.

The preparation took four times as long as the meeting. None of it required her expertise. All of it required her time.

The workflow in this issue runs that preparation automatically, twice a day, so that by the time a meeting appears on her calendar, the agenda is already built, the context is already pulled, and the deck is already generated.

Five tools. One pipeline. Runs without you.

Step 1: Set up Day.ai as your data layer. Day.ai functions as an AI operating system that aggregates data from your emails, past meetings, and text messages into one searchable source. Before anything else in this workflow can run, Day.ai needs to be connected to your communication channels. Think of it as the memory layer: every email thread, every client conversation, every meeting transcript lives here and becomes queryable by the rest of the system. Setup time: 30 to 60 minutes to connect your accounts and let it sync existing history.

Step 2: Connect Todoist as your task layer. Todoist is where all action items, deliverables, due dates, and assignees live. The workflow uses it as the single source of truth for what needs to happen and when. If you already use a task manager, you can adapt this workflow to your tool. Todoist works well here because it has a clean API that Claude can write to automatically.

Step 3: Build the daily context sync skill in Claude. This skill runs twice a day, at 7 a.m. and 3:35 p.m. Use this Claude prompt as the foundation:

"You are a task extraction assistant. Scan the following emails, texts, and missed calls from the past 12 hours: [Day.ai data input]. Identify any outstanding deliverables, commitments made, or action items with implied deadlines. Format each as a Todoist task with: task name, due date if mentioned or implied, assignee if clear, and a one-line note on the source. Output in Todoist-compatible format."

Schedule this skill to run automatically through Claude Cowork or a scheduling tool like Zapier. It keeps your task list current without manual entry.

Step 4: Build the hourly agenda scan. This skill runs every hour, checks your calendar for meetings in the next 24 hours, and builds a structured agenda for each one. Use this prompt structure:

"Check the calendar for meetings in the next 24 hours. For each meeting found, search Day.ai for relevant email threads, past meeting notes, and open tasks related to the attendees or project. Generate a meeting agenda in markdown format that includes: BLUF (the single most important decision or outcome needed from this meeting), historical context (two to three sentences on relevant background from previous communications), prioritized action items using the RICE method (Reach, Impact, Confidence, Effort), and specific items each attendee needs to prepare or decide. Save the agenda as a markdown file in Day.ai."

The output is a complete, context-rich agenda generated from your actual communication history, not from memory.

Step 5: Generate the presentation with Gamma. Take the markdown agenda file and run it through Gamma's Aurora engine with your company style guide uploaded. Gamma transforms the raw agenda into a branded visual deck automatically, matching your fonts, colors, and layout preferences.

The deck is ready before you open your laptop for the meeting.

1. Use Prompt Cowboy before building any skill. The quality of every automated skill in this workflow depends on the quality of the prompt behind it. Prompt Cowboy takes a rough description of what you want and refines it into a prompt that works consistently without constant tweaking. Before building any new skill, run the concept through Prompt Cowboy first. It saves hours of iteration later.

2. The BLUF is the most important line in the agenda. Every meeting should have one sentence at the top that answers: what decision needs to be made, or what outcome needs to happen, for this meeting to be worth the time? If the automated system can't generate a clear BLUF from the available context, that's a signal the meeting may not need to happen at all. Use the absence of a BLUF as a filter.

3. The RICE method turns a task list into a priority list. Most task lists are ordered by recency, not importance. The RICE scoring in the agenda skill (Reach, Impact, Confidence, Effort) gives every action item a calculated priority so the meeting focuses on what matters most, not what was added last. Run RICE on your Todoist backlog quarterly using Claude and watch how differently your week gets organized.

4. Start with one meeting type before automating everything. Don't try to run this system on every meeting from day one. Pick one recurring meeting, a weekly client update or a Monday team standup, and build the workflow around that specific context. Once it runs cleanly on one meeting type, expand to others. A system that works on one meeting is more valuable than a system that half-works on ten.

5. The presentation isn't the point. The preparation is. Gamma's deck is the visible output, but the real value is everything that happened before it: the context pulled from Day.ai, the tasks updated in Todoist, the RICE prioritization. The deck is just proof that the system ran. Even without the visual output, a team that walks into every meeting with a BLUF and three prioritized action items makes better decisions faster.

Beyond the Bid Circle is a peer community for builders scaling with systems and AI—the same builders who've escaped the 60-hour grind. Swap war stories, shortcuts, and real strategies with people who get it.