How should leaders adjust their approach to planning tools given the shift toward a context layer?

Leaders should stop treating planning tools like Linear or Salesforce as the sole source of strategic context. Instead, they should create lightweight, versioned artifacts, like Markdown files, that capture learnings, bets, and decisions. Scheduled agents can sync these artifacts with execution tools, ensuring both humans and AI agents have access to the same up-to-date context.

What practical steps can teams take to avoid rebuilding context from scratch every quarter?

Teams can maintain a living set of artifacts that capture key learnings, updated assumptions, and major decisions as they happen. By keeping these documents short, explicit, and regularly refreshed, quarterly planning becomes a review and integration point rather than a reset. This approach helps teams stay aligned and reduces cognitive overload.

How does the idea of context engineering change the role of annual and quarterly planning rituals?

Annual and quarterly planning shift from producing static plans to creating shared context that guides decision making. Annual planning becomes a time to codify learnings and big bets into durable artifacts, while quarterly planning focuses on integrating new insights and reaffirming tradeoffs. The emphasis moves from rigid forecasting to maintaining clarity and adaptability as conditions change.

Strategic Planning is Essentially Context Engineering

For years, building a business was constrained by how fast teams could execute.

So we built planning systems around that constraint. Annual plans in decks. Quarterly priorities in planning tools. Roadmaps in product software. Forecasts in spreadsheets. Decisions scattered across meetings and Slack.

It worked because humans had time to stitch it all together.

That world is disappearing (and fast).

Strategic planning is no longer just the process of setting goals and reviewing progress. It is the process of creating, maintaining, and distributing the context that humans and agents use to make good decisions.

Strategic planning is essentially context engineering.

Annual planning is for memory and bets

Annual planning still matters because leadership needs a ritual for turning experience into doctrine.

What did we actually learn last year?
Which assumptions held up?
Which bets worked?
Which did not?

Those lessons do not magically turn into shared understanding. Assuming everyone walked away with the same insights is naive. Everyone has their own account of everything that’s ever happened. I have been in plenty of sessions when someone else's account of a situation was nothing like mine.

Annual planning forces us to codify these lessons into shared context.

The annual planning cycle should produce a small set of durable artifacts:

What we learned and what we believe now
The big bets we are making
The non-goals protecting focus
High-level revenue and investment posture for the year

That is the point of annual planning. Not false precision. Memory and bets.

Quarterly planning is for alignment, review, and integration

Quarterly planning matters too, but it should not feel like rebuilding context from scratch every 90 days.

If a company is running a continuous planning motion, the quarter becomes a cross-functional review and integration point.

Start with what changed. What did we learn? Which assumptions still hold, and which need updating? Reaffirm the key tradeoffs. Set clear asks for product, engineering, and commercial teams for the next 90 days.

That is much better than treating quarterly planning like a reset. And much better for your teams’ cognition of your business strategy.

Real-time planning is about keeping context current

This is where the model becomes practical.

A company stays aligned when its context layer is up to date.

Within the quarter, the rolling wave of planning tightens. The next 30 days should be clear and ready for execution. The following 30 days should be shaped enough for prototyping and risk reduction. The final 30 days remain more conceptual and flexible.

Set your roadmaps out as far as you need, but obsessing over perfect requirements is wasted effort. The world is moving too fast for that now. Clarity matters, but rigidity costs more than it returns.

To be clear: Continuous planning does not mean constantly changing priorities. It means the next set of work becomes clearer as execution approaches. And it happens through the regular rhythm of the business: prioritization, discovery, delivery and retrospectives.

The result: An operating system that stays in sync with reality.

When context is maintained in real time, annual and quarterly planning both improve. Companies can set targets (E.g. OKRs) with higher confidence. Quarterly planning focuses on review and integration, not rebuilding.

Planning tools do not go away

This does not mean companies stop using Linear, HubSpot, or their finance tools.

They should not.

Those systems are still the systems of record for execution and state. Linear is where product development moves. HubSpot or Salesforce is where commercial reality lives. Finance is still going to run in whatever finance system the company trusts.

The mistake is making those tools the only place context exists.

Those systems are very good at tracking status. They are much worse at being the shared context layer for a company where humans and agents both need to understand what is happening and why.

That is why I think the right model is simple:

Tools remain the systems of record for state. Markdown becomes the system of record for context. Scheduled agents keep the two in sync.

Not bi-directional magic. Not a giant new platform on day one. Just a practical sync loop.

A scheduled task can pull from Linear to update the roadmap context. Another can pull from HubSpot or finance systems to refresh the revenue plan. Another can update GTM assumptions before a weekly sales pipeline review.

Humans still define meaning. Agents keep the context current.

A practical system

A simple version of this can look like:

Markdown

strategy/
   2026/
      annual-plan.md
      annual-learnings.md
      quarterly-plan/
         previous-quarter.md
         revenue-plan.md
         roadmap.md
         gtm-plan.md
         decisions/
            dec-014-pricing.md
            dec-015-launch-scope.md

Each file should be short, current, and explicit.

annual-plan.md captures the year’s bets, non-goals, and resource posture.

annual-learnings.md captures the key learnings from the previous year.

previous-quarter.md shows what we accomplished and learned in the previous quarterly “step.”

revenue-plan.md is shaped by finance and GTM, but can be refreshed from CRM and financial systems.

roadmap.md is shaped by product and engineering, but can be refreshed from Linear (or Jira, etc.) on a schedule that lines up with cycle delivery.

gtm-plan.md captures launch assumptions, campaigns, and commercial dependencies.

The decisions/ folder holds the choices that would otherwise disappear into chat and meetings.

That is enough structure for humans to navigate and agents to use.

Where this breaks down

GitHub or something equivalent can work as a starting point because it gives you versioning, review, and history. But it is obviously not perfect for a multi-disciplinary company. Most non-technical teams do not want to live in Github.

That is fine. It is still a useful substrate for a first version.

But I suspect this points to a real gap in the market. We do not really have a great operating system for human-agent businesses yet. We have systems of record. We have collaboration tools. We have planning tools. What we do not really have is a clean company-wide context layer that is easy for every function to use and easy for agents to work against.

That probably gets invented.* Or vibe-coded internally first.

*I’d bet good money the foundational model companies are building this. Anthropic is the obvious candidate given their success in B2B

Either way, that is a tooling gap. It is not the main point.

The main point is that the company needs the context layer whether the perfect tool exists yet or not.

The shift

In a human-only company, strategy could survive as a collection of presentations, dashboards, and tacit understanding.

In a human-agent company, that is no longer good enough.

The companies that adapt best will not be the ones with the prettiest planning dashboards. They will be the ones that treat strategy as a shared, versioned, continuously refreshed context layer.

Annual planning still matters because companies need memory and bets.

Quarterly planning still matters because teams need alignment, review, and integration.

Real-time planning matters because context must stay fresh.

But the output of those motions should increasingly live in artifacts that humans can read, agents can use, and scheduled workflows can keep current.

In a world of humans and agents, strategy is not just a plan. It is the context layer of the business.

A different take on the “AI-native organization”

There is a growing view that AI will fundamentally reshape the org chart.

You can see it in how leaders like Jack Dorsey are thinking. The idea is that companies become smaller, flatter, and increasingly coordinated by AI systems rather than layers of management.

That direction is probably right in some ways. But I think it misses the deeper constraint. Even in an AI-native company, someone still has to define:

What matters
What tradeoffs exist
What decisions have been made
What the system is trying to optimize for

AI can coordinate work. It cannot invent coherent context on its own. (Yet?).

The real bottleneck is not organizational structure, it is whether the company has a shared, usable context layer that both humans and agents can operate against.

That is why I think the evolution of strategic planning matters more than the evolution of the org chart.