The standard pattern with strategic plans is depressingly consistent. You spend two weeks in November writing the plan. The OKRs land in a doc that everyone agrees with in the offsite. By February, the doc is open in nobody's tab. By April, you're operating from instinct and reacting to the inbound, and the OKRs are something you'll dust off for the board update because the board asks. The plan isn't wrong. It's just structurally disconnected from the working week.
The fix isn't a more elaborate planning ritual. It's holding the plan, the rationale behind each goal, and the weekly signal that tells you whether you're tracking — all in one vault where the agent can read across them and surface the drift before it becomes a quarter you have to explain. The same shape underwrites annual planning and goal setting at the cycle start and investor-relations and board updates at the cycle's regular checkpoints.
A vault shaped around the plan, not the deck
The shape that holds up across plan complexity is roughly: one top-level page for strategic planning, with sub-pages for the current plan, the rationale archive, the OKR tracker, the weekly-signal journal, the quarterly-review log, and the prior-cycles archive. The current plan sub-page is the canonical living document — short, rewritten as the plan actually evolves, not maintained in parallel with a deck nobody opens.
Capy supports unlimited page nesting, so each OKR can fan out into its own sub-page with the underlying initiatives, the leading indicators, and the working notes from the team or function that owns it. The whole vault is plain markdown. That matters because every Monday morning, you can ask the agent to read across the plan, the rationale, the OKR tracker, and the past week's signal, and write a one-paragraph orientation: what's tracking, what's drifting, what needs attention this week.
OKRs in a database that lives inside the plan
Most teams track OKRs in a spreadsheet or a dedicated tool that drifts out of sync with the actual working week within a month. The structured tool is fine; it's just structurally separate from where the work happens, which means it gets updated quarterly under duress instead of weekly under habit.
A working alternative: an inline OKR database directly inside the plan page via the :::database::: directive — rows for objective, key result, current value, target value, owner, last updated, status. The database lives directly inside the page, not in a separate tab. Each row links to a sub-page with the underlying initiatives, the leading indicators, and the per-week signal.
The OKR review stops being a context-switch into a different tool and starts being a five-minute pass through the page you're already in. Ask the agent weekly to read the OKR database and surface anything where the current value hasn't moved in three weeks, anything where the trend is in the wrong direction, anything where the owner hasn't logged any signal. The drift gets caught early instead of in the quarterly post-mortem. (The same compounding-database discipline is the spine of contract negotiation with AI notes — different artifact, identical mechanic.)
Rationale captured at the moment, not reconstructed later
The plan that drifts is rarely the one whose goals were wrong. It's the one whose rationale gets lost. Why did we pick this objective and not the other one. Why did we set the target at this level. Why did we deprioritize the initiative the team kept asking about. By March, the rationale is gone, and "we set this in November" is the only answer left, which doesn't help when the world has changed.
A working setup: a rationale sub-page per major objective with three short sections — why this objective now, what we considered and rejected, what would make us reconsider. Three to five paragraphs total. Written in the planning week, not after the fact.
When the plan needs to flex mid-cycle — and it always does — you read the rationale before changing direction. Either the original rationale still holds (don't change just because the inbound is loud) or it doesn't (change explicitly, and update the rationale sub-page with what changed). The plan stops being either rigid or random. It becomes a conversation.
A weekly signal journal that keeps the plan honest
The structural reason plans get disconnected from the work is that nobody's writing down the weekly signal that bears on whether the plan is right. The customer who churned and named a specific reason. The competitor announcement that affects the why-now. The hire who didn't pan out. The sales meeting where a prospect raised something nobody on the team had thought about.
A working setup: a weekly-signal journal sub-page where you drop one-line entries throughout the week as signal lands. Two minutes per entry. By Friday, the week's signal is captured, even if it's just bullet points.
Each Friday, ask the agent to read the week's signal and the current plan, and to surface anything in the signal that bears on a plan assumption — anything that supports an assumption, anything that contradicts one, anything that suggests an assumption nobody named in the planning week is showing up in the work. The plan stays connected to the actual customers and competitors and hires. (The capture-while-it's-fresh habit is the same one underwriting first-time founders moving faster — different setting, identical mechanic.)
Quarterly reviews that don't restart from blank
Most quarterly OKR reviews are a panic. The team scrambles to score each KR, write commentary, and produce a slide deck for the board, all in the same week. The review has the shape of accountability theater because the actual reflection time gets squeezed by the assembly time.
A working flow: by the time you sit down for the quarterly review, the agent has access to the full quarter's signal journal, the OKR database with weekly updates, and the rationale pages. Ask the agent to read across all of it and draft a one-page review per objective: how we're tracking against the KRs, what the signal suggests about whether the rationale still holds, what we'd change for next quarter, what we'd carry forward. You edit. The team review meeting becomes a working session against an actual document.
The board-facing version then becomes a curation pass over the working version, in the team's actual voice. (The drafting-from-vault pattern is the same one underwriting investor-relations and board updates — different audience, same speed-up.)
Prior cycles that teach the next cycle what to be
Most teams write each year's plan as if the prior years didn't exist. The patterns from last year's misses don't make it into this year's plan because the prior plan and its post-mortem live in a Drive folder nobody opens.
A working setup: each prior cycle's plan, OKR final scores, and end-of-cycle retrospective live as plain markdown in the prior-cycles archive. When you sit down to write the next year's plan, ask the agent to read the prior three cycles and surface the patterns: which objectives consistently land below target, which key results were structurally too easy or too hard, which assumptions kept turning out to be wrong, which initiatives kept getting started and abandoned.
You're not constrained by the prior cycles; you're informed by them. The plan that lands has a much better chance of being calibrated to your team's actual track record. (The same lessons-archive discipline is the spine of documenting lessons learned after every project — different scale, identical compounding.)
The planning offsite is the one cycle event that usually warrants recording — twelve hours of conversation across two days, dozens of decisions and rejected directions. Record in Capy where the protocol allows; the transcript comes back with speaker diarization (labels like "Speaker 1: …") so you can tell who said what. Ask the agent to draft a recap with three sections per objective: what we agreed, what was raised but not resolved, what assumptions we made. When somebody asks in March "wait, did we discuss the alternative on objective two," you find the moment with the speaker labeled. (The recording-with-speaker-labels habit shows up in our writeup of coaching sessions and personal growth — different setting, same loss pattern.)
What this isn't
Capy isn't an OKR platform. The structured side — if your team uses one — still lives there. Capy is for the unstructured side: the rationale, the weekly signal, the quarterly review, the prior-cycles archive. That's the part that's currently sprawled across decks and Slack and your memory, and the part that determines whether the plan stays alive or quietly dies.
It's also single-user by design. One operator, one vault. If your team needs a multi-user shared planning workspace where the leadership team edits the same artifacts with role-based permissions, that isn't this product. The shape that fits is the founder, ED, or operating leader running the personal connective layer alongside whatever shared tools the team uses for the structured side. Pricing tiers are on the pricing page.
A small first test
Pull out last year's plan. Drop it on a Capy page along with the actual end-of-year results (paste them in or attach the export) and any signal you can recall about why the misses happened. Ask the agent to draft a one-page retrospective: which objectives missed and why, which assumptions turned out to be wrong, what you'd change about next year's plan. If the retrospective surfaces a pattern you'd otherwise have rolled into next year's plan unaddressed, that's the agent doing for you what strategic planning actually deserves.
Try Docapybara free. Load one cycle's plan and one quarter of signal and see what falls out.