LifeOS Review 2026: One AI Dashboard for Everything

What LifeOS bundles

LifeOS is Daniel Miessler's project, and the repository at github.com/danielmiessler/LifeOS shows 19,614 stars. The pitch on the site and in the repo is a universal AI harness: goals, context and agents in one system, aimed at moving a person from current state to ideal state in work and life. That framing matters for a review, because it is not the same product category as a task manager with an AI button bolted on.

The bundle is tasks, notes, calendar and an AI assistant. The question worth asking is how much of that is native and how much is a thin wrapper over somebody else's API. The documentation describes the assistant as the center of the system, with goals and context feeding it, and the surrounding productivity surfaces as the places where that context gets captured and acted on. Calendar integration is described as a connection to an external calendar provider rather than a homegrown scheduling engine. That is a wrapper, and it is a reasonable one, but it means calendar behavior is bounded by whatever the upstream provider allows.

Tasks and notes read as native in the docs. They are stored in the system's own structure, which is why the AI can reason across them. If tasks were proxied to Todoist and notes proxied to Notion, cross-source reasoning would be a federation problem, and the docs do not describe it that way. The repository structure supports this: the harness and its context files are the substance, and the productivity surfaces exist to feed them.

Pricing starts at $0/mo. The site lists a free tier and paid tiers above it, with an annual discount on the paid plans. I would treat the free tier as the honest evaluation path, because the whole value proposition depends on whether the AI layer actually helps you, and that is not something a pricing page can tell you.

One thing the bundle does not include, and this is worth stating plainly: LifeOS is not a replacement for a document editor, a spreadsheet, or a project management tool with dependencies and gantt views. It is a personal system. Teams looking for a shared workspace will find the model wrong for them.

Setup and data import

Import is where consolidation products live or die. The docs list importers for Notion, Todoist and Google Calendar. What survives and what mangles is the interesting part, and the documentation is more candid than most about the shape of the problem.

Notion import is the messiest of the three. Notion pages are block trees with nested databases, and any importer has to flatten that into something the destination can hold. The docs describe the import as bringing over page content and converting databases into the system's own structured format. What does not survive cleanly is the relational layer: Notion relations and rollups depend on Notion's query engine, and a flattened copy loses the live link between records. If your Notion workspace is mostly prose, import is close to lossless. If it is mostly relational databases, expect to rebuild the relations by hand.

Todoist import is cleaner because tasks are a simpler shape. Projects, tasks, due dates and labels map onto the destination's equivalents. Subtasks and recurring rules are the usual casualties in any task import, and the docs do not promise perfect fidelity there. Comments and attachments are the other common gap.

Google Calendar import is a subscription or sync rather than a one-time copy, which is the right call. Duplicating calendar events into a second store creates drift the moment either side changes. The docs describe the connection as ongoing, so the calendar stays where it is and LifeOS reads it.

Setup itself is account creation plus connecting the AI provider. The repository shows the harness expects an API key for a model provider, which is a meaningful detail: you are not buying inference from LifeOS, you are bringing your own. That keeps the subscription price low and pushes model cost onto you. At current rates, a mid-tier model like openai/gpt-5.5-pro at $30/M input and $180/M output, or anthropic/claude-opus-4.1 at $15/M input and $75/M output, a heavy personal user will spend real money on tokens. A light user will spend very little. The pricing page does not bundle inference, so budget for it separately.

The import path I would recommend: start with the calendar connection, import Todoist next because it is the least lossy, and treat Notion as a selective migration rather than a bulk one. Pull the pages you actually reference and leave the archive where it is.

The AI layer: useful or decoration

Three capabilities matter here: prompt-to-task, weekly review, and search across sources.

Prompt-to-task is the entry point. You type something like "remind me to follow up with the accountant about the Q3 filing next Tuesday" and the system is supposed to produce a task with a due date. The docs describe this as a core flow. Whether it works depends on the model you connect and how much context the harness supplies. Because you bring your own key, the quality ceiling is your model choice, not LifeOS's. That is a genuine advantage over tools that hide the model behind a subscription and cap your usage.

Weekly review is the feature that distinguishes LifeOS from a generic assistant. The docs describe a structured review that pulls goals, recent activity and open items into a single pass. This is where the goals-and-context framing pays off: a review that knows what you said you wanted six weeks ago can point out that you have not touched it. A review that only sees this week's tasks cannot.

Search across sources is the consolidation payoff. One query, results from tasks, notes and calendar. The docs describe it as unified. The honest caveat is that unified search is only as good as the index, and the index is only as good as the import. If half your Notion workspace stayed in Notion, search will not find it, and you will be back to two search boxes.

On hit rate: I have not run a scripted query set against this, and I am not going to invent one. What I can say is that the architecture predicts where it will fail. Queries about things you captured inside LifeOS should hit. Queries about things that live in a connected calendar should hit if the connection is live. Queries that require reasoning across a large imported corpus will degrade in proportion to how much the import mangled.

The decoration risk is real but specific. If you connect a cheap model and never write goals or context, the assistant becomes a chat box with your task list attached. The system's value is a function of the context you feed it. That is a design choice, not a flaw, but it means the product is worse for lazy users than a simpler tool would be.

Export and lock-in

Export is the test most reviews skip, and it is the one that determines whether the subscription is a decision or a trap.

The docs describe a full export of your data. The format is structured, and the repository being open means the schema is inspectable rather than a black box. That is a meaningful difference from closed competitors: even if the export format is awkward, you can read the code that produces it and write a converter.

Fidelity on the way out has the same shape as fidelity on the way in. Tasks and notes export cleanly because they are native. Calendar events do not export at all in a meaningful sense, because they were never LifeOS's data; they live in Google Calendar and stay there. That is fine. The AI context, the goals, the accumulated history of reviews, is the part that is genuinely yours and genuinely at risk if export is incomplete. The docs describe goals and context as exportable. I would verify that on day one rather than year three.

Time to rebuild elsewhere is the number that matters. For a tasks-and-notes corpus, rebuilding in Notion or Obsidian is a weekend of scripting against a structured export. For a system where the value is the accumulated context and review history, rebuilding means losing the thing that made it valuable. That is the real lock-in, and it is softer than a data hostage situation but harder to escape than a CSV.

My read: the open repository is the strongest anti-lock-in signal here. A closed product can change its export policy in a release note. An open one cannot quietly delete the exporter without someone noticing.

LifeOS vs Notion AI vs Khoj

These three are not the same product, and the comparison table should be read with that in mind.

• Price. LifeOS starts at $0/mo with paid tiers above and an annual discount; inference is billed by your model provider. Notion AI is priced per seat on top of a Notion plan. Khoj has a free tier and paid tiers, and can also be self-hosted at no license cost.
• Self-host. Khoj is the clear winner; the project is built to run on your own hardware. LifeOS is open source and the repository is public, so self-hosting is possible in principle, but the docs describe a hosted product as the primary path. Notion AI is closed and hosted only.
• Offline mode. None of the three is a true offline-first product. Khoj self-hosted on a local machine gets closest. LifeOS and Notion AI both assume connectivity for the AI layer.
• Mobile. Notion has the most mature mobile apps of the three. LifeOS has mobile apps; app store ratings and recent review themes are worth checking before you commit, because mobile parity is where consolidation products most often fall short of their desktop experience. Khoj's mobile story is thinner.
• AI quality. LifeOS and Khoj both let you bring your own model, which puts the ceiling in your hands. Notion AI is a fixed model behind a subscription. For a personal system where context matters more than raw model power, LifeOS's goals-and-context framing is the more interesting design; for document Q&A over an existing Notion workspace, Notion AI has the shorter path because the data is already there.

If you want a self-hosted personal AI over your own notes, Khoj is the more direct fit. If you want an agent-oriented workspace with a plugin ecosystem, LobeHub is worth a look. If you want a system that treats your life goals as first-class data, LifeOS is the one making that bet.

Beetlix is our own product. Where the comparison is fair: Beetlix and LifeOS both assume you bring model access rather than bundling inference into a flat subscription, so both put model cost on you and both let you pick the model. Beyond that the products diverge; LifeOS is a personal life-management harness, and I would not present them as substitutes.

Privacy and sync

Where your data sits is the question the privacy policy answers, and the answer here is layered.

Your tasks, notes and goals sit in LifeOS's hosted store on the paid and free hosted tiers. Your calendar stays with Google and is read over a connection. Your AI prompts and the context sent with them go to whichever model provider you configured. That third leg is the one people forget: the privacy policy covers LifeOS's handling, but the model provider's policy governs what happens to the text of your prompts. If you connect a provider with weak retention terms, your notes are subject to those terms the moment they enter a prompt.

The docs describe encryption in transit and at rest for stored data. That is table stakes and I would not treat it as a differentiator. What the policy does not claim, and no hosted product of this shape can claim, is end-to-end encryption with server-side AI: if the server has to read your notes to answer a question, the server can read your notes.

Telemetry is the clause to read closely. The policy describes collection of usage data for product improvement, which is standard. The distinction that matters is whether note content is included in that telemetry or only interaction metadata. The policy language separates them, and I would take that at face value while noting that policies change and the open repository is your early warning system when they do.

Sync behavior across devices is described as automatic. Conflicts are the practical concern: two devices editing the same note offline will produce a merge, and the docs do not promise a specific conflict resolution strategy. For a personal system used by one person, this rarely bites. For anyone sharing an account, it will.

Verdict: buy, wait, or skip

Five axes, scored honestly.

• Consolidation value: 4/5. Tasks, notes, calendar and AI in one place, with the AI actually reading across them. The calendar is a wrapper, but a live one.
• AI usefulness: 3.5/5. The design is right and the model ceiling is yours to raise. It rewards context input and punishes neglect, which is fair but demanding.
• Import and export: 3.5/5. Todoist and calendar are clean. Notion is lossy on relations. Export is structured and the open repo makes it trustworthy.
• Mobile and offline: 3/5. Mobile exists but is not the strength. No true offline mode for the AI layer.
• Price and lock-in: 4/5. Free tier, annual discount on paid, bring-your-own-model keeps the subscription honest, open repository limits lock-in.

Overall: 3.6 out of 5.

Buy if you are a single person running your own life system, you already pay for a model API, and you are willing to write down your goals and context so the assistant has something to work with. The free tier is a real evaluation path, and the annual discount makes the paid tier reasonable once you know it fits.

Wait if your working life is mostly relational Notion databases, or you need mobile parity before you commit, or you are not prepared to spend on tokens. The import will frustrate you and the assistant will underperform until you feed it.

Skip if you need a team workspace, a true offline-first tool, or self-hosting as a hard requirement. Khoj is the better fit for the last one.

Price per value: at $0/mo to start, the only real cost of trying is your time on import. That is a low enough bar that the question is not whether to look, but whether the goals-and-context model matches how you actually think about your work. For a certain kind of person it will, and for that person the consolidation is worth the switching cost. For everyone else, the four-tool stack is fine.

How this review was researched

This review draws on the vendor documentation at ourlifeos.ai, the official pricing page, the public repository at github.com/danielmiessler/LifeOS, and live AI model pricing data for the providers listed. No hands-on testing was performed; claims about behavior are attributed to the documentation and repository, and claims about cost are attributed to the pricing pages and the live pricing snapshot.