The honest reason most annual plans don't survive past March is that they were written without enough context. You sit down in late December, you've been heads-down for months, and the only inputs to the plan are whatever you can pull from memory plus a couple of dashboards. The goals come out aspirational, the priorities come out reasonable-sounding, and three months later you can't remember why exactly you committed to that number on that line.
A plan that holds up is one that's written from your actual year, not from a January self-image. AI notes earn their keep here when they hold the year — board memos, weekly reviews, retrospectives, customer notes, hiring decisions, the deals that closed and the ones that didn't — in one place where the agent can read across all of it and produce an honest read of what happened. Then the plan gets written from that read. If your year also runs through a board-of-directors meeting cycle, the same vault feeds both the quarterly board prep and the year-in-review.
A planning vault that holds the whole year
The shape that holds up is roughly: one top-level page per year, with sub-pages for the plan itself, the quarterly reviews, the running list of decisions made, the goals scoreboard, and any board or advisor memos. Below that, the rest of your operating notes — weekly reviews, customer conversations, hiring debriefs — live in their normal places in the vault. The point isn't to centralize everything; it's that the agent can read across all of it on demand.
Capy supports unlimited page nesting, so a planning page can fan out into per-pillar or per-team sub-pages without you committing to a depth in advance. The plan, the goals scoreboard, and the running decisions log are all plain markdown notes the agent can read and edit directly.
Inline databases help here. A goals scoreboard with rows for goal, owner, target, current, and status lives directly in the plan page via the :::database::: directive. You don't switch tabs to a separate tracker. Quarterly, the agent can read the scoreboard and write a one-paragraph summary of where things stand — which is the kind of housekeeping that's easy to skip and expensive to skip.
Start with an honest read of last year
The hardest part of annual planning isn't deciding what to do next. It's writing an unflinching read of what actually happened last year. Most founders skip this step because writing it from scratch is depressing and slow, and the version they write is too polished to be useful for planning.
Drop your monthly reviews, board memos, and quarterly retros from last year on a single page. Tell Capy: read across these and write a one-page honest summary of the year — what we said we'd do, what we actually did, where the variance was, what the biggest surprises were, what we learned about our own decision-making. The agent has every entry; it doesn't need to round off the rough parts.
The first draft is rarely perfect; it's usually a much better starting point than the polished retro you'd write from memory. You edit, you push back on parts that feel wrong, you ask the agent for the supporting passages from the underlying notes when you want to verify a claim. By the end of an hour you have a read of last year that's grounded in the actual record.
Draft the new plan from the honest read, not from a template
Once you have the read of last year, the new plan gets a lot easier to write. You're not staring at a blank page; you're answering specific questions that the read raised.
Tell the agent: based on the year-in-review summary, draft a v1 of next year's plan with three pillars, two or three goals per pillar, and a one-paragraph rationale per goal that explicitly references what worked or didn't work last year. The draft is yours to rewrite, but it inherits the honesty of the read. The goals tend to come out more specific because they're anchored to actual variance you saw, not to a generic OKR shape.
The same flow works for personal annual plans, team plans, and product roadmaps. The pattern is the same: read the year, then write the plan from the read.
Customer and team conversations as planning fuel
The richest planning input most founders don't use is the body of customer and team conversations they had during the year. The customer call where someone said the magic words about why they bought. The team retro where three people independently surfaced the same drag on velocity. These signals usually live in scattered notes and rarely make it into the annual plan.
Record those conversations inside Capy when you have them. The transcripts come back with speaker diarization — labels like "Speaker 1: …" so you can tell who said what. The recordings sit with their summaries on the relevant page.
When you're planning, ask the agent: pull every customer conversation in the last six months where the customer said something that sounded like a positioning insight, and summarize the patterns. Or: pull every team-retro transcript and find the dragging-on-velocity comments that came up more than twice. The agent reads across the transcripts and writes the answer. That's the kind of synthesis that's almost impossible to do honestly from memory.
Goals that get revisited because revisiting is cheap
The reason most annual goals decay isn't that they were bad goals. It's that the cost of revisiting them — finding the doc, re-reading the rationale, checking against current reality — was high enough that nobody did it past February.
A working setup makes revisiting cheap. The goals scoreboard lives in the same vault as your weekly reviews. Once a month, ask the agent to read the recent weekly reviews, compare them to the goals scoreboard, and write a one-paragraph status that flags any goal that's drifting and any goal whose original rationale is starting to look wrong. You're not running a status meeting with yourself; you're getting a structured read in the time it takes to drink coffee.
Quarterly, do a heavier version of the same thing. Ask the agent to draft a quarterly review using the structure of the one you wrote last quarter. Edit, share with whoever needs to see it, file it in the planning page. Next year's annual review will read across all four of these and the year-in-review will already be half-written.
The decisions log that makes the next plan easier
The single highest-leverage planning artifact most founders don't keep is a decisions log. A short page with one entry per material decision: what was decided, what the alternatives were, what the rationale was, what we expected to happen. The same log discipline is the backbone of AI notes for co-founders: alignment, decisions, and accountability — different audience, same compounding artifact.
The reason to keep one isn't paperwork. It's that next year's plan gets dramatically easier to write when you can read across the year's decisions and see the pattern. The places where you keep making the same kind of bet, the places where your stated rationale and your actual behavior diverge, the bets that are still pending verdicts — all of that is invisible without the log and obvious with it.
Ask the agent to help you keep it: after a notable conversation or working session, point it at the relevant notes and ask it to propose an entry. You confirm or edit; the agent writes it in. The cost of keeping the log gets close to zero, which is the only way it actually gets kept.
What this isn't
Capy isn't a goals-tracking SaaS. It doesn't have a Monday-morning email cadence or a manager view. The structured-tracker side of goal management still belongs in the tools that handle that. Capy is for the unstructured side of planning — the year-in-review writing, the rationale, the running decisions log, the synthesis across customer and team conversations — which is the part that actually drives whether the plan is honest.
It's also single-user by design. One person, one vault. If your goal is a team workspace where every direct report has visibility into the master plan with role-based permissions, that isn't this. The shape that fits the founder doing personal and company-level planning is a vault that lives with the person making the calls.
A small first test
Take last year's plan and last year's monthly reviews and load them on a single page in Capy. Ask the agent to write a one-page year-in-review using the structure described above. If the read names a tension between what you said you'd do and what you actually did — one you'd been carrying around without saying out loud — you've got a sense of what the agent does for next year's plan when the year is half-done. Working through a personal year too? Our writeup of career transitions and job searches with a vault behind you covers the personal-life version of the same review.
Try Docapybara free. Load last year's reviews and see what the year-in-review looks like.