Five jobs for your agent, every morning.

You get a short list of genuinely unpublished drafts ranked by how close they are to done, with the top pick named and a brief reason why.

The catch: The agent marks a draft as unpublished when it is already live, because the site search missed a match.

Before your first meeting you have one short brief per external meeting, with sources for every claim about each guest.

The catch: The brief describes the wrong person with the same name.

You have a folder of interlinked markdown pages with an index and a change log, built from your sources, that you can read in Obsidian and query through the agent, with every change reviewable in git.

The catch: Changelog entries say one thing while the page's main text still describes the old state.

You get a local dashboard of spending by month and category that you can rebuild each month, with totals that match your statements.

The catch: The model labels most transactions as "other" or picks odd categories, such as fuel for bank fees.

You get a structured P&L with each transaction classified, totals calculated with formulas, and a clear view of gross margin.

The catch: The agent misclassifies unusual Amazon fee types such as storage surcharges, reimbursements, or co-op charges.