A daily ledger that knows why each item is on it.

A trustworthy daily surface — manually composed each morning from your routines, your active objectives, and the follow-ups you captured yesterday.

Every commitment lives in one place, connected to the reason it matters. Tasks attach to concerns. Concerns ladder to horizons. Horizons serve intended situations. When you check something off, you're not reducing a list — you're advancing something that matters.

Items rolled forward carry their provenance — when they moved, why. The system remembers. Changes are data, not errors.

Today composes from the substrate. Routines that are due. Tasks anchored to today's date. Playbook steps awaiting human review. Items rolled forward from previous days. Action mode for execution; planning mode for composition. Date navigation for past or future days. Items can be edited inline; metadata can be added with the elbow-arrow primitive (concern, horizon, time block, source).

From this hour to this decade — one cascade.

Time is not a calendar. It's a stack of horizons — century, decade, year, quarter, month, week, day — each scoping what matters at that scale.

Multi-scale planning without losing context. Move freely between zoom levels. The structure of work persists; the time mapping is flexible. Replan without restructuring objectives. Same objective structure, different timeframes — work decomposes by function, by time, or by concern, your choice.

Three-panel cascade — parent, current, child. Six horizon scales (Century → Day) plus an All mode. Concern-based filtering. Coherence indicators showing strategic alignment between adjacent scales. Items can carry horizon badges; rescheduled items show their original temporal placement alongside their current one.

Everything you're responsible for, in one tree.

A clean hierarchy of what's on your plate — the company you're building, the role you're advancing in, the body you're maintaining, the people you're investing in.

The complete map of your responsibilities. Operational coherence — the harmony across domains through deliberate active and inactive states. Permission to say no without guilt; the inactive concerns are still preserved, just not invested in. Six months of operating context in one place — because you put it there once, and the system kept it organized.

This is the substrate. Every action you take updates the same context an agent will later read from.

Hierarchical tree with unlimited nesting. Each concern carries: objectives, active experiments, attached sources, documents, philosophy memos, hours-per-week allocation, and history. Active / dormant states. Investment signals visible per concern (experiments running, routines, hours allocated). The Universal Concern Context lets you scope every other page to a single concern or multi-select across branches.

Hypotheses with a deadline.

When you don't know whether something will work, you don't add it to your routines. You run an experiment.

Behavioral archaeology. Over months and quarters, the experiment record reveals patterns — which kinds of hypotheses you tend to commit to, which domains keep failing, what your win rate looks like. Self-knowledge without judgment. Not aspirational tracking — actual data on how you operate.

Each experiment is attached to a concern with a hypothesis, a protocol, and an evaluation date. Three-question completion: did I follow the plan, did I commit the time, did I get the result. Won experiments promote to routines. Lost experiments record the lesson and the constraint discovered. Source links connect experiments to the worldview that informed them.

Systems that run when you do.

Daily, weekly, quarterly, annual. Designed once, then executed without re-inventing.

Your operating cadence, not someone else's playbook. The hardest thing about operating a business — or a life — isn't building the systems. It's running them consistently. Routines are how you stop making the same decisions every Monday. Decision fatigue compounds; the routine eliminates it.

Recalibration is preserved. A target that moves from 3×/wk to 2×/wk keeps the previous value. The sequence of calibrations is itself the data — the process of finding real operating parameters.

Recurrence types: specific days, daily, weekly, monthly, frequency-based (N times per week/month/year). Custom schemas per routine — track distance and pace for runs, weight and reps for lifts, food items and calories for meals. Skip with reason. Active / paused states. Adherence shown as scorecard rows on the Today page and aggregated in the Routine page history.

Same playbook. Different actor.

Today, you run them yourself — every step. Tomorrow, the routine steps run against your context overnight, and you wake up to drafts and decisions waiting at the human checkpoint.

A playbook is an operator's proven system — a structured protocol that can be run manually today and agent-assisted tomorrow. Same concept at different automation levels. Your weekly outbound sequence. Your hiring loop. Your monthly board prep. Your training cycle.

As agents commoditize execution, your playbooks compound. The methodology you've designed runs against the same context whether you, your team, or an agent is the actor in the loop.

A playbook is a sequence of steps. Each step has an actor (you, your team, an agent, or a human-review checkpoint), inputs, outputs, and structured fields. Steps can be triggered manually, on schedule, or by upstream events. The same workflow step list renders a marathon-training playbook for a person and an outbound playbook for an SDR — only the actor indicators change. Agent steps write back to the substrate with provenance.

Horizons isn't a chatbot. It isn't a generic workflow automation tool. It's the structured operating context that makes both human and agentic action safe, useful, and accountable. Agents don't float around the app. They run approved playbook steps against known context, with provenance and human-review checkpoints.

The trace becomes the record.

Where you intended your week to go, where it actually went, and what to do about the gap.

Most operators don't lose to bad strategy. They lose to a week that quietly eats their strategic hours and replaces them with reactive ones. Time Budget makes that visible. The variance is the data — not a judgment, just an honest record of how strategy translated to time.

Time is the binding constraint. Every life has 168 hours per week. Most planning tools hide this constraint behind priorities and due dates; Horizons elevates it. You cannot do everything, so allocation is the strategy.

Hours allocated by concern at the start of the week, with rollup from children to parents. Auto-population from routines and experiments. Forward-looking budget plus backward-looking accounting. The Default Week — a declarative weekly shape placing concerns on weekday/weekend slots. Push-only Google Calendar export. Variance shown as statement-table rows; categories that flex (1:1s, external meetings) get neutral indicators; categories that signal trouble (strategic underrun, reactive overrun) get warning treatment.

Read the world. Build the synthesis.

Sources you trust — books, podcasts, channels, newsletters — distilled into a working synthesis you can actually operate from.

Knowledge organized by concern relevance, not by source. Resurfaced when you're working on the relevant concern. The 80/20 of the books and podcasts you've consumed, structured into operating implications you can act on — not a library you have to maintain.

Three-tier hierarchy. Artifact — individual pieces of content (a YouTube transcript, a book chapter, a newsletter issue), each with a one-pager summary. Source — synthesized knowledge base per author, accumulated across all artifacts regardless of medium. Domain — cross-source synthesis by topic, producing a unified framework integrating lessons from multiple sources. Sources are polymorphic — one author may produce videos, books, and a newsletter; the synthesis is medium-agnostic.

An operator's command room, on one operating graph.

Eight surfaces. Five pillars. One source of truth. Built individual-first, architected for operator teams.

Horizons gives the operator a simple daily work surface while preserving the larger architecture underneath. You sign in and see only what you need to do today. But every item ladders to a concern, an intended situation, a project, an operating outcome, a metric, a playbook, a document, a source, or a time budget when you zoom out.

The product is broad because the problem is broad. The operator's coherence problem cannot be solved by a narrow task app, a narrow strategy app, a narrow project manager, a narrow AI memory layer, or a narrow automation canvas. Strip any pillar and the gap returns.

Underneath all eight surfaces is the same operating graph: intended situations, concerns, time horizons, time budgets, tasks, projects, routines, documents, metrics, decisions, observations, sources, playbooks, agents, and provenance. The graph is what makes Horizons more than the sum of its surfaces — and what makes it the context layer for human and agentic work alike.

The substrate is the product. The agent is an actor.

Most AI tools are chatbots — they hold context in a conversation, lose it, hallucinate, depend on whatever you've pasted in the last thirty minutes. The substrate problem is the actual problem.

Horizons exposes structured operating context to agents through MCP — intended situations, concerns, constraints, decisions, documents, metrics, provenance, approval boundaries. Not a chat history. Not a vector store of pasted snippets. The actual graph of intentions, work, decisions, documents, metrics, and provenance that you've built by using the app.

When your agent needs context, it queries the same substrate you read from. When it learns something, it writes back with provenance. Both doors open.

Playbooks make agent actions accountable. Each step has an actor. Each actor has constraints. Each output writes back to the substrate with provenance. Where judgment is required, the playbook stops at a checkpoint and waits for you.

Today the steps run with you in the loop. Tomorrow, the routine steps run overnight, and you wake up to drafts and decisions waiting at the human checkpoint.

The same workflow step list renders a 12-week marathon training playbook (you the actor), an SDR's outbound week (you the actor), an agentic SDR's outbound run (agent the actor), and a competitive-intelligence agentic workflow (agent the actor). Only the actor indicators change.

As agents take on more of the routine steps, the substrate doesn't change. The methodology you've designed runs against the same context. The provenance accumulates. The judgment checkpoints stay where you put them.

Horizons playbooks are exportable as skills — the format that's becoming a cross-industry standard for AI agent capabilities. Design methodology in Horizons; export as a skill any agent can invoke. Skills compound. Prompts evaporate.

A language system for operator-grade work.

Horizons is not just software. It's a language system. Before you have language for something, you cannot quite see it; before you can see it, you cannot act intentionally about it.

These phrases differentiate Horizons in the market. They're not just describing the product — they're constituting what the product is.

Three tiers. One operating instrument.

Horizons charges premium for individual operators, more for agent-equipped operators, meaningfully for operator-led teams. The architecture is the same; the seat count, agent budget, and team capabilities scale.

Direct Horizons stays paid and premium. The price filters for serious users, signals the value of system thinking, and funds the operator-grade build standard.

Operating field notes.

Longform writing on the practice of running a serious life — what works, what doesn't, what's compounding, what isn't.

Posts go up here as they're written. The intent isn't a content marketing channel; it's a place to think out loud about the discipline of running an Operator OS — and to share what shows up in my own data.

— On behavioral archaeology: what twelve months of routine adherence actually showed me.
— Same playbook, different actor: notes from running an agentic SDR loop overnight.
— The middle (months 4–7): why most goals die there and what survives the dip.
— Inactive isn't failure: a quarter of deliberately parked concerns and what came back.

First posts publish later this year. Subscribe via RSS or sign up for the operating brief.