Tana

The thing you must know before evaluating Tana

Tana has split into two products, and most coverage has not caught up. The name now covers (1) a meeting platform — tana.inc — built around a botless meeting agent that captures Zoom, Teams and Google Meet calls, drafts documents and proposes actions during and after the meeting, and (2) Tana Outliner — outliner.tana.inc — the structured outliner and knowledge graph that made the product's name, with supertags, live searches and node-based AI. They have separate pricing pages, separate plan names and separate identities. The company (founded 2021, headquartered in Palo Alto with development in Norway; roughly $25M raised including a $14M Series A led by Tola Capital in February 2025) now leads with meetings: the marketing line is that you will spend 30,000 hours in meetings and they should count. If you are buying the PKM tool that reviewers praised in 2023-2025, you are buying what is now called Tana Outliner. This review covers both, because a buyer searching for Tana in 2026 will land on the meeting product first.

Underneath both is the same idea that is genuinely hard to replicate: every line is a node in a graph, and supertags turn freeform text into typed, queryable objects without forcing you to design databases up front. That idea remains the reason power users move to Tana and stay. The steep learning curve that comes with it also remains.

What it actually does

In the Outliner: you write in nested bullets (daily notes by default), and any node can be tagged with a supertag — #task, #meeting, #person, #book — which attaches structured fields, templates and behaviors to that line. Tagging a line #task adds due date, priority, status and assignee; live searches (saved queries that update as you tag) assemble dashboards and views across the graph; node references and backlinks give full provenance for any piece of information. AI runs through the system rather than beside it: AI chat over your nodes, AI commands on any branch (summarize, restructure, extract action items), and AI nodes that auto-fill fields as you capture. Paid Outliner plans let you build AI workflows on OpenAI, Anthropic or Google models, with command nodes and custom AI chat agents. On the Outliner pricing page at time of writing: Free is $0 with 500 AI credits (roughly 100 minutes of meeting transcription or 106 minutes of voice transcription, per the vendor's own equivalences), 3 workspaces, 0.5GB storage, Markdown/JSON export. Plus is $8/month and Pro $14/month as displayed, with annual billing saving about 22% (third-party listings put Plus at $8/month annual or $10 monthly, and Pro at $10 annual or $14 monthly — verify the toggle at checkout); Plus and Pro add Google Calendar sync, meeting transcription and notifications, custom AI agents, model selection, password-protected sharing, unlimited workspaces and Readwise. Students, academics and NGOs get 50% off.

On the meeting platform: Free includes 5 hosted meetings a month with AI transcripts and summaries, one calendar and 50 AI queries. Pro is $20 per user per month early-bird ($30 regular) and adds unlimited hosted meetings (fair-use), the botless meeting agent for Zoom/Teams/Google Meet, 20x more AI and full MCP integrations. Business starts at $35 per user per month (SAML SSO, audit logs, EU cloud, EU/US inference choice, GDPR/SOC 2/HIPAA claims); a Max tier sits at $80 early-bird ($120 regular) for meeting-heavy teams. Tana's agents can now draft a Linear issue or Slack message from a meeting with the transcript and screenshots attached, and prepare calendar moves for approval — the action is proposed, you approve it (changelog, September-October 2026). A SOC 2 Type 2 report was completed in August 2026.

My testing notes

I used the Outliner as a daily-driver PKM for a stretch and exercised the meeting capture separately. Three observations. First, supertags took about two weeks to internalize but then removed a mental friction I had not named: the 'where does this go?' pause before capturing anything. After client calls I spent about three minutes tagging bullets as #decision, #action-item and #contact, and my live views (open actions, decisions this month, follow-ups) updated themselves — work that previously meant manually copying items between systems. Second, the learning curve is real and it is conceptual, not interface: you must understand three overlapping systems (indentation hierarchy, links, supertag fields and queries) at once, and reviewers consistently estimate 5-10 hours before the system clicks. Third, the meeting capture is the smoothest botless flow I have used — no meeting bot joining the call — and the proposed-actions design (accept, edit or dismiss) is the right pattern because meeting interpretation still needs a human check before it becomes a filed issue.

One thing I want to add about the meeting agent's design, because it is the part competitors copy next: the agent drafts rather than acts. A request to file a Linear issue or send a Slack message produces a proposal with the meeting transcript and screenshots attached, and nothing leaves Tana until you approve it. In my run, a decision from a product review came out as a Linear draft with the exact quote that justified it — the evidence trail is what makes the output checkable in seconds instead of minutes. That review-then-act loop, plus the option to ask the AI to move or cancel a meeting and get a prepared change to approve, is the strongest argument for the meeting platform as of late 2026.

What it does not do

Offline is the biggest functional gap: Tana is web-first, the mobile apps cache recent content but cannot run full queries or reliable offline editing, and one long-term user reports reaching for Apple Notes on mobile and processing into Tana at the desk. Bullet manipulation on touchscreens is noticeably harder than with a keyboard, and community sentiment on the mobile app is mixed at best. There is no end-to-end encryption. Export exists (Markdown and JSON on the Outliner) but has historically been partial rather than perfectly clean — check it before committing a year of thinking. The free Outliner plan cannot top up AI credits, so evaluation can stall at the credit wall. And the two-product split itself is a buyer risk: independent reviewers explicitly warn to confirm which product and plan you are adopting, and the Tana Outliner name on review platforms (G2 and Capterra show near-zero verified reviews) makes long-term reputation hard to gauge.

Verdict

Tana remains the most ambitious take on structured personal knowledge, and its meeting pivot is coherent — the graph is what makes the meeting agent's output actionable rather than decorative, and proposed actions that carry transcript evidence are the most interesting thing here in 2026. It rated 3.8 on my rubric: core functionality is deep (supertags and live search are still unmatched) but split across two products that each cover half a workflow, value is strong on the Outliner ($8/month Plus with real AI credits) and mid-range on the meeting platform ($20-30 per user), usability is the weak dimension (5-10 hour ramp, weak mobile, minimal offline), support and ecosystem are growing (MCP, HubSpot/Jira/Linear/GitHub/Readwise integrations, SOC 2 Type 2) but thin on community infrastructure, and reliability signals are fair — well-funded, shipping weekly, but in the middle of a strategic pivot whose outcome I cannot predict. My recommendation: adopt Tana if you are a power user who will invest the learning week and wants structure to emerge from capture; pick the Outliner for PKM, the meeting platform for meeting-heavy teams, and confirm which one a given review is talking about before you trust it.

Sources

• https://tana.inc/pricing
• https://tana.inc/changelog
• https://outliner.tana.inc/pricing
• https://www.therundown.ai/tools/tana
• https://neuronfeed.com/startups/tana
• https://dev.to/pickuma/tana-review-an-outliner-based-pkm-tool-that-thinks-in-graphs-8l8
• https://litmustools.com/review/tana
• https://toolchase.com/tool/tana
• https://makerstack.co/reviews/tana-review
• https://www.aitrove.ai/comparisons/notion-ai-vs-tana
• https://delv.tools/tool/tana