Taskade Review 2026: An AI Workspace That Builds Apps, and What the Credits Really Cost
Taskade started as a real-time outliner with projects, docs and mind maps. In 2026 it sells itself as Taskade Genesis: describe a business tool in plain English and get a live app with its own data, AI agents and automations. This review is for small teams deciding whether to run their work there. It covers the plans and seats, how AI credits are spent and what happens when they run out, the refund terms in Taskade's exact words, where your AI requests are processed, and how it compares with Notion AI, ClickUp, Lindy and Relevance AI.
There are no affiliate links in this review. Every Taskade link goes straight to taskade.com or its documentation.
| Made by | Taskcade Inc., San Francisco (YC S19, founded 2017) |
| What it is | Projects, docs and mind maps, plus AI agents, automations and prompt-built apps (Taskade Genesis) |
| Paid plans | Pro $10/mo yearly ($20 monthly), 10 users; Business $25 ($50), unlimited users |
| Meter | AI credits: 50,000/mo on Pro, 150,000 on Business |
| Free tier | Yes: 2 users, 6,000 one-time credits |
The best price-per-person in this category, wrapped around a credit meter and a contract you should read first.
"For a team of three to ten people, Taskade is hard to beat on sticker price: one $10-a-month Pro plan, billed yearly, covers ten users, agents, automations and unlimited apps. What you are really buying is AI credits, and Taskade's terms say every charge is final."
Here is how I read Taskade in October 2026. The product has been renamed in all but the logo. Taskade's own press kit now says to write "Taskade Genesis" for the product and "Taskade" for the company, and the pricing page opens with "Build AI apps, agents, and automations in real time." The outliner, task lists, docs and mind maps are all still there, and they are still free. But the money, and the risk, now sit in the AI layer.
The pricing model is unusual and, for teams, generous. Seats are bundled, not sold per person: Free covers 2 users, Pro covers 10, and Business, Max and Enterprise cover unlimited users. Taskade's billing FAQ puts it plainly: "AI usage is metered in credits, not seats." So a team of eight pays the same $10 a month (yearly) as a solo founder, and both draw from the same 50,000 monthly credits.
Three things temper that. First, credits are spent per request, and the cost depends on the model: Taskade's docs price Claude Sonnet 4.6 at 110 credits per request, and its own guide shows Claude Opus 4.6 at 400. Second, the Terms of Service (updated Sep 30, 2026) say "Payments are final," with only three narrow exceptions. Third, Taskade's security page is unusually candid that it has "no SOC 2 report to hand you yet" and does not sign a Data Processing Agreement with customers, while its privacy policy says some model providers it routes to are "based outside the United States, including in China."
Not a hands-on test: JAVIS did not sign up for, build on or pay for Taskade for this review. It is built from Taskade's pricing page (both billing toggles), help center and documentation, Terms of Service, Privacy Policy and Security page, its changelog and blog, its official YouTube channel, the US App Store and Google Play listings, competitors' pricing pages, and public reviews on Trustpilot, Product Hunt and Hacker News, all as of Oct 4, 2026.
Projects remember, agents reason, automations run, and Genesis wraps them into an app.
Taskade calls its architecture "Workspace DNA": Memory (projects and databases), Intelligence (AI agents) and Execution (automations). Taskade EVE is the chat agent you talk to, and TSK-1 is the routing layer that picks a model. Under the branding, this is how the pieces fit, and which of them spend credits.
A small YC company that pivoted a nine-year-old workspace into an AI app builder, and ships almost daily.
Y Combinator S19, per Taskade's own blog. We found no later round on taskade.com.
Taskade's footer claim. Registered users, not paying customers.
Genesis apps created, per Taskade's Sep 30, 2026 post. Not paying apps.
Changelog count; latest entry Sep 25, 2026.
Taskade now positions itself against several groups at once. Its homepage compares Genesis with code generators (Lovable, Bolt, Cursor, v0), AI chat apps (ChatGPT, Claude, Gemini), project tools (Notion, Linear, Asana) and automation tools (Zapier, n8n, Make), and its footer links to comparison pages for Notion, ClickUp, Lindy and Relevance AI. That breadth is the pitch: one workspace instead of four subscriptions. It is also the risk, because a small team is competing on every front.
| When | What happened | Source |
|---|---|---|
| 2017 | Founded by John Xie (CEO), Dionis Loire and Stan Chang (CTO) | About page |
| Oct 24, 2019 | $5M seed; Y Combinator Summer 2019 | Taskade blog |
| Dec 12, 2022 | Taskade AI first reads a project | Taskade blog |
| May 29, 2023 | AI Agents launch in beta | Taskade blog |
| Oct 10, 2025 | Taskade Genesis launches | Taskade blog |
| Jul 8, 2026 | TSK-1, the "Taskade System Kernel," routes work across 15+ models | Taskade blog |
| Sep 30, 2026 | Terms, Privacy Policy and Security page all updated | taskade.com |
| Name you will see | What it actually is |
|---|---|
| Taskade Genesis | The product name now, and the prompt-to-app builder inside it |
| Taskade EVE | The chat agent that builds apps, agents and automations for you |
| TSK-1 | Model routing, not a model: "It is not another model" |
| TSK-Instant to TSK-Reasoning | Four depth tiers in the model picker, plus TSK-Auto |
| Workspace DNA | Projects + agents + automations working as one loop |
| App Kits, Community Gallery | Cloneable apps; creators can sell paid kits |
People love how fast an idea becomes an app. The complaints are about credits, changed deals and support.
Each source as shown on Oct 4, 2026, reported on its own. App Store figures are for the US storefront (app "Taskade: AI Apps & Agents," seller Taskcade Inc., version 6.0.12 of Sep 24, 2026); Google Play figures are for the US store (1M+ downloads, updated Sep 15, 2026). The two stores disagree by almost a full star. Taskade's own pricing FAQ claims "4.9/5 rating from 5,000+ reviews" without naming the source; neither US store matches it. We did not include G2, Capterra or Reddit.
"The credit limits on the pro plan don't give you much leeway into creating personalized stuff ... I ran out of AI tokens in the pro plan with small features requested"
"At some point, my monthly credits dropped from 50k to 25k with no warning or communication from Taskade."
"The way it seamlessly combines tasks, notes, mind maps, and chat into one unified workspace is a true game-changer for productivity."
| Pattern | Where it shows up | What it means for you |
|---|---|---|
| Idea to working app fast, no code needed | Trustpilot's review summary; Product Hunt | A real strength for teams with no developer |
| Credits run out sooner than expected | Product Hunt (Pro plan); Trustpilot, Apr 30, 2026 ("Credits are King, and Kings are Pricey") | Model your credit use before you pay yearly |
| Lifetime or legacy plans lost unlimited AI | Trustpilot, Jul 17 and Jul 27, 2026 | Taskade's billing FAQ confirms the change was "on purpose" |
| Support slow to answer | Trustpilot, Aug 22, 2026 ("written to Taskade support four times"); profile marked "Hasn't replied to negative reviews" | Do not count on fast help for a broken automation |
| Mobile got harder after the AI expansion | Product Hunt | Test on phones if your team works there |
| Pro bundles 10 users even for a solo user | Product Hunt ("10 imaginary users") | Great for teams, no solo discount |
Four paid plans, priced per workspace rather than per person, with yearly billing at half the monthly rate.
Free
Free forever, no card.
- 2 users, 1 workspace
- 6,000 one-time AI credits
- 3 apps, 1 agent, 3 automations
- Lighter AI models only
Pro
Billed yearly ($120). $20 a month on monthly billing.
- 10 users included
- 50,000 AI credits a month
- Unlimited apps, agents, automations
- 100+ integrations, API, MCP
Business
Billed yearly ($300). $50 a month on monthly billing.
- Unlimited users
- 150,000 AI credits a month
- Custom domains, white-label
- SAML SSO, admin controls
Max
Billed yearly ($1,200). $200 a month on monthly billing.
- Unlimited users and teams
- 400,000 AI credits a month
- Autonomous agents
- Extended thinking, per-agent models
Swipe the table sideways to see every column.
| Plan | Yearly (per month) | Charged per year | Monthly billing | Users | AI credits |
|---|---|---|---|---|---|
| Free | $0 | $0 | $0 | 2 | 6,000 one-time |
| Pro | $10 | $120 | $20 | 10 | 50,000 a month |
| Business | $25 | $300 | $50 | Unlimited | 150,000 a month |
| Max | $100 | $1,200 | $200 | Unlimited | 400,000 a month |
| Enterprise | "Custom," Contact sales (see note) | Unlimited | 1.1M a month | ||
| Extra credits | "$5 /pack," on top of plan credits; pack sizes shown only when signed in | Pack credits "never expire on paid plans" | |||
Swipe the table sideways to see every column.
| Feature (pricing page table) | Free | Pro | Business | Max | Enterprise |
|---|---|---|---|---|---|
| AI apps / agents / automations | 3 / 1 / 3 | Unlimited | Unlimited | Unlimited | Unlimited |
| AI knowledge base | 250 MB | 100 GB | 1 TB | 1 TB | 1 TB |
| GPT, Claude and other premium models | No | Yes | Yes | Yes | Yes |
| 100+ integrations, API, MCP | No | Yes | Yes | Yes | Yes |
| Remove "Powered by Taskade" | No | Yes | Yes | Yes | Yes |
| Custom domain, white-label, SAML SSO | No | No | Yes | Yes | Yes |
| Autonomous agents, extended thinking | No | No | No | Yes | Yes |
| Bring your own key (BYOK), SCIM | No | No | No | No | Yes |
Not an affiliate link. It opens Taskade's own pricing page.
Your apps keep running when credits hit zero. Your AI stops, and you cannot overdraw.
This is the part Taskade handles better than many rivals. Its pricing FAQ says a published app keeps serving visitors with no credits left, and a build cannot run past your balance. What it does not give you up front is a price list: the per-model cost lives in the in-app model picker, and the docs warn that "published figures drift as models and pricing change." Here are the rules, in Taskade's words.
| Rule | Exact wording (source) | What it means |
|---|---|---|
| What costs credits | "Credits power all AI work in Taskade: generating apps, running agents, and executing AI steps in automations." (Pricing FAQ) | Non-AI work is free |
| Cost per request | "Claude Sonnet 4.6 is priced at 110 credits per request, for example." (Credits & Billing docs) | Model choice drives the bill |
| Same prompt, different cost | "A long, complex build can cost several times a short one." (Pricing FAQ) | Hard to predict per build |
| Monthly reset | "Your monthly credits reset on your billing anchor date." (Billing FAQ) | Unused plan credits do not carry over |
| Bought credits | "Purchased and earned Usage credits do not expire." (Model credits guide) | Packs are bankable |
| Hitting zero | "Your credit balance is a hard ceiling ... generation stops the moment that balance is used up." (Pricing FAQ) | No surprise overage |
| Apps at zero | "Your apps never stop. Published apps, projects, collaboration, real-time sync ... and non-AI automations keep working normally." (Pricing FAQ) | The business tool stays up |
| Agents answering your users | "An agent answering end-users ... This is the line item that grows as your app gets traffic." (Credits & Billing docs) | Visitors are free; their AI chats are not |
| Auto top-up cap | "The cap is effectively controlled by how often you hit that threshold." (Credits & Billing docs) | There is no monthly spend limit |
No refund window, no money-back guarantee, and no lifetime deal for new buyers.
Some older guides and search snippets still mention a 30-day satisfaction guarantee. Taskade's current Terms of Service, updated Sep 30, 2026, say the opposite, and the help-center cancellation article (updated Mar 18, 2026) only offers to "investigate" a charge made in error. Here is the exact wording.
| Topic | Exact wording (Terms of Service, Sep 30, 2026) |
|---|---|
| Refunds | "Payments are final. Except as set out below, all charges are non-refundable. This includes subscription fees, renewal charges, seat charges, AI credit purchases, and paid App Kit purchases, and we do not issue refunds or credits for partial billing periods, unused AI credits, or unused seats." |
| The only exceptions | "(a) where a refund is required by applicable law, including consumer rights you cannot waive; (b) where you purchased through the App Store or Google Play, in which case that store's refund process applies to that purchase; and (c) where we terminate your paid plan without cause" |
| Billing errors | Contact support "within 30 days of the charge ... This billing-error review is not a refund period and does not create a right to a refund for charges that were correctly made." |
| Cancelling | "Your Paid Account stays active for the remainder of the billing period you have already paid for, and reverts to the Free plan at the end of that period. Cancelling does not retroactively refund a charge that has already been made." |
| AI credits | "no cash value, are non-transferable, and are non-refundable once purchased" |
| After deletion | "we may delete your content and have no obligation to retain it, so export anything you need before closing your account." |
The pricing page offers no lifetime deal, and the legacy Starter plan is "not offered to new buyers." For people who bought AppSumo or similar deals, the billing FAQ says: "LTDs were sold before Taskade AI existed. Taskade extended AI credits voluntarily as a goodwill gesture. They are not part of the original LTD purchase." Lifetime tiers keep unlimited workspaces, users, projects and storage, and can use their own OpenAI, Anthropic or Google key.
Price changes apply only "at your next billing cycle after we notify you, never mid-cycle." Downgrades take effect at renewal. A failed renewal payment gets "a short grace window." And when you cancel, you keep your projects on the Free plan.
What 50,000 credits buys, and how Taskade's team price compares.
Taskade does not publish a full per-model price list, so the table below uses the two prices it does publish: 110 credits for a Claude Sonnet 4.6 request (docs) and 400 for Claude Opus 4.6 (in its usage guide). Treat these as rough ceilings: a single agent run can be several requests, because "each reasoning turn and tool call is a separate step."
| Monthly credits | Sonnet 4.6 requests (110 each) | Opus 4.6 requests (400 each) |
|---|---|---|
| Free: 6,000 one-time | About 54 (Free uses lighter models) | Not available on Free |
| Pro: 50,000 | About 450 | About 125 |
| Business: 150,000 | About 1,360 | About 375 |
| Max: 400,000 | About 3,630 | About 1,000 |
Swipe the table sideways to see every column.
| Monthly sticker price | Taskade | Notion | ClickUp + Brain AI | Lindy | Relevance AI |
|---|---|---|---|---|---|
| 5 people | $10 (Pro) | $100 (Business) | $80 | $149.95 | $234 (Team) |
| 12 people | $25 (Business) | $240 (Business) | $192 | $359.88 | $234 (Team) |
| How it is priced | Per workspace | Per member | Per user, plan + AI add-on | Per active user | Per plan, builder seats |
Decide by team size, how much AI you will run, and whether anyone will ask for a security review.
Start on Free. Lists, boards, tables, calendars, docs and mind maps are free; the 6,000 credits cover a first app.
Pro. Try it monthly ($20) first, then yearly ($10 a month) once credit use is clear.
Business ($25 yearly, $50 monthly). Custom domains, white-label and SSO start here.
Not today. Taskade says it has no SOC 2 report yet, signs no DPA, and has no HIPAA BAA. Look at Notion Enterprise, ClickUp Enterprise or Lindy Enterprise.
Lindy fits that shape better. For sales-ops agent pipelines, Relevance AI.
Describe a business tool, get a live app with its own data, agents and automations, then publish it.
Genesis is what makes Taskade different from a project tool with a chatbot. You type something like "a CRM that tracks my leads and follows up automatically," and EVE builds the app's projects (the data), one or more agents, and the automations around them. The pricing FAQ lists the kinds of things teams build: dashboards, client portals, CRMs, booking systems, storefronts, forms and internal tools. Published apps can have their own sign-in for your users, Stripe checkout and, on Business and up, your own domain.
Taskade's 4.5-minute walkthrough, published Sep 2, 2026 on the channel linked from taskade.com. It shows one prompt turning into an app with agents and automations wired in, which is the exact workflow this section describes.
| Picker option | Taskade's description | Credit cost |
|---|---|---|
| TSK-Auto | "Recommended · picks the best model" | Varies with the task |
| TSK-Instant | "Fast edits" | Cheapest |
| TSK-Standard | "Everyday tasks" | Mid |
| TSK-Thinking | "Multi-agent planning" | Higher |
| TSK-Reasoning | "Workflows and systems" | Highest |
| Pinned model | "pin one model, cost on hover" | Shown per model |
How portable is a Genesis app? You can export an app to GitHub through a branch-and-pull-request flow. Taskade's docs describe the bundle as capturing "everything needed to reconstruct the app in another workspace," meaning another Taskade workspace: projects, agent definitions, automations, media and branding. API tokens, connected-account credentials and your app users' accounts are not included. Treat it as a backup and version history, not as a way to self-host the app elsewhere.
From a prompt to a branded portal your clients log into, with a human checking what the agent sends.
Here is the job Taskade is best placed for: a five-person agency that tracks client work in Taskade projects and wants clients to see their own status without another tool. The diagram marks where AI acts, where credits are spent, and where a person approves.
The approval step is real but opt-in: Taskade's agent-tools docs describe "Human in the Loop" as "An approve/reject gate before agents interact with external platforms," which you switch on per tool. The docs suggest using it "during the first few weeks of agent deployment." For anything client-facing, I would keep it on.
The original Taskade is still here, and it is still the cheapest part to use.
Every Genesis app is built on ordinary Taskade projects, which means the same data can be a list, a board, a table, a calendar, a mind map or a doc. That shared data model is what lets an agent read your task list and an automation update it. It is also why long-time users still pick Taskade as a lighter alternative to Notion for outlining and mind mapping.
| Project view | Free | Paid plans |
|---|---|---|
| List, Board, Table, Calendar | Yes | Yes |
| Mind Map, Docs | Yes | Yes |
| Gantt Chart (with Timeline) | No | Yes |
| Actionsheet, Org Chart | No | Yes |
| Unlimited projects | Yes | Yes |
Agents learn from your files and projects, act through tools, and run on automations that retry.
A Taskade agent has knowledge (files, URLs, YouTube, or live projects), tools (built-in ones such as web search plus integrations like Slack, Gmail and HubSpot), memory, and a model. AI Teams put several agents together so the best-qualified one answers or they hand work to each other. Automations are trigger-and-action flows with branching, looping and retries; agents can call them, and they can call agents.
| Limit | Free | Pro and up |
|---|---|---|
| AI agents | 1 | Unlimited |
| AI automations | 3 | Unlimited |
| Automation runs | 10 a month | "1,000 per seat a month" on Pro |
| Premium connectors (Gmail, Slack, Stripe, Salesforce, Notion, Airtable, HubSpot) | No | Yes |
| HTTP webhook triggers | No | Yes |
| Autonomous agents, extended thinking | No | Max and up |
Taskade's Sep 24, 2026 changelog adds a useful guard: turning an automation on now checks for unfinished steps first, and a step that does not match a filter shows as "filtered, not failed." An "Automation Health" card in settings flags failed automations "before they auto-pause."
100+ integrations inside, and an MCP server so Claude or Cursor can work in your Taskade from outside.
Taskade connects to Slack, Gmail, Google Workspace, GitHub, Stripe, HubSpot, Salesforce, Notion, Airtable, WhatsApp, Zapier and Make, among others. The free plan "runs the standard connectors"; the premium set needs a paid plan. On every paid plan Taskade also runs a hosted MCP server, so an AI client can read and update your projects, run automations, and hand work to a Taskade agent.
| Question | Taskade's answer |
|---|---|
| Which plans? | "MCP server access is included on every paid Taskade plan" |
| How do clients sign in? | "OAuth 2.0 (PKCE)"; agents that run without a browser use a token from Settings > API |
| Does it cost credits? | Yes: "Each action an external MCP client runs through the hosted MCP server" (AI usage guide) |
| Which clients? | Claude Desktop, Cursor, Claude Code, Windsurf, Zed, Inkeep, LangGraph |
A one-minute Taskade video, published Sep 27, 2026, on what you can wire up once an app is live: analytics, your own domain and integrations.
You pay Taskade credits for every model, and only Enterprise and old lifetime tiers can bring their own key.
Unlike several agent platforms, Taskade does not pass model costs through at a published rate. You buy credits, and each model spends a different number per request. The upside is simplicity: "You do not need your own API keys." The downside is that you cannot check the markup, and you cannot move heavy use onto your own OpenAI or Anthropic bill unless you are on Enterprise.
| Bring-your-own-key question | Taskade's answer |
|---|---|
| Who can use BYOK? | Enterprise ("Bring your own OpenAI and Anthropic API keys"), plus AppSumo and PitchGround lifetime tiers |
| What does a key cover? | "your custom agents," and with an OpenAI key, built-in automation steps set to TSK-Standard or TSK-Reasoning |
| What does it not cover? | "It does not cover Taskade Genesis app building or Taskade EVE chat" |
| The catch | "A model left on Auto still uses Taskade credits." |
Where your AI requests go. Taskade's privacy policy (Sep 30, 2026) is specific: "The model you pick decides who processes your request." Its example provider list includes "OpenAI, Anthropic, Google (including Vertex AI), and Amazon Web Services (including Bedrock)" and "Alibaba (Qwen), DeepSeek, Moonshot AI (Kimi), Z.ai (GLM), MiniMax, xAI (Grok), and NVIDIA." It adds that "Some of these providers are based outside the United States, including in China," and that on Auto, "your plan's default routing picks one for you." Taskade will send "the current list of the AI providers your account can be routed to" if you email support.
Good basics and SSO from Business up, but no SOC 2 report, no DPA, and no redaction of sensitive data.
Taskade's security page, updated Sep 30, 2026, is one of the more honest in this category: it says plainly what is not done yet. Here are its claims in its own words, so you can hand them to whoever approves your tools.
| Topic | Exact wording | Source |
|---|---|---|
| SOC 2 | "An independent SOC 2 audit is part of the program and is not complete. We have no SOC 2 report to hand you yet" | Security page |
| DPA and SCCs | "We do not currently offer a signed Data Processing Agreement or Standard Contractual Clauses to customers." | Security page |
| Sensitive data | "We do not scan for or automatically redact regulated data." | Security page |
| HIPAA | "We have no HIPAA Business Associate Agreement" | Privacy Policy |
| Training | "Taskade does not use your workspace content to train generalized AI models" | Terms |
| AI logs | Kept "normally no more than 30 days" | Privacy Policy |
| Encryption | "TLS on every connection"; "AES-256 encryption for backups and stored credentials" | Security page |
| Hosting | "Single AWS region (US East)"; recovery targets of under 4 hours (RTO) and under 1 hour (RPO), "not contractual commitments" | Security page |
| Certifications | GDPR "full compliance in progress"; ISO 27001 "certification in progress"; Google CASA for its Google API scopes | Security page |
| Control | Free and Pro | Business and Max | Enterprise |
|---|---|---|---|
| Roles (Owner to Viewer), workspace isolation | Yes | Yes | Yes |
| SAML SSO (Okta, Azure AD, Google) | No | Yes | Yes |
| SCIM provisioning | No | No | Yes |
| Bring your own model keys | No | No | Yes |
| Human approval before agent actions | Per tool | Per tool | Per tool |
| Risk for a small team | Taskade control | What you still need to do |
|---|---|---|
| An agent emails a client something wrong | "Human in the Loop" per tool | Turn it on for every outbound tool |
| Client data reaches a provider you did not approve | Model choice decides the provider | Pin approved models; avoid Auto for sensitive agents |
| Prompt injection in an incoming email or form | Prompts "sanitized"; "No filter stops every attempt" | Keep write actions behind approval |
| A shared Editor link leaks | Regenerating the invite link now kills the old one (Sep 25, 2026) | Use email invites with Viewer by default |
| Auto Top-Up runs up a bill | Threshold setting only | Keep thresholds low; review the Usage page weekly |
Best for small, non-technical teams that want one workspace with AI, and do not face a vendor security review.
| Team and job | Why Taskade fits or not | Likely plan |
|---|---|---|
| 3 to 10 people running projects, docs and a few agents | One flat price covers everyone; workspace features free | Pro |
| Agency building client portals and dashboards | Genesis, App Users, custom domains, white-label | Business |
| Founder building internal tools with no developer | Prompt-to-app plus automations and Stripe | Pro, with credit packs in build sprints |
| Team that lives in outlines and mind maps | The original strength; mostly free | Free or Pro |
| Company that needs SOC 2, a DPA or HIPAA | Not offered today | Look elsewhere |
| Engineers who want code they own and host | Exports rebuild inside Taskade, not a standalone codebase | Consider a code-first builder |
Eight real drawbacks, each traced to a Taskade page or a user report.
No refund window or guarantee in the Sep 30, 2026 terms, including for unused credits and seats.
Per-request costs live in the in-app picker; the docs publish only a Sonnet 4.6 example (110 credits).
Taskade says both plainly. That rules it out for many client and enterprise reviews.
Depending on the model, including on Auto routing, per the privacy policy.
Lifetime and legacy users lost unlimited AI or saw allowances cut; terms allow rates to change again.
Only a threshold. A busy public agent can keep buying packs.
Enterprise "$250/mo" vs "Custom," AI Teams on Pro or Business, free-tier credit packs yes or no.
Unanswered tickets on Trustpilot and a profile marked "Hasn't replied to negative reviews."
Flat team pricing, a real free plan, apps that keep running at zero credits, no overdraw, honest security disclosures, and very fast shipping.
Pick by what you need more: a workspace, a project manager, or a dedicated agent platform.
Swipe the table sideways to see every column.
| Product | Best when you need | Entry price (Oct 4, 2026) | How AI is metered | Our review |
|---|---|---|---|---|
| Notion AI | A polished docs-and-wiki workspace with agents, and enterprise controls | Free; Plus $10, Business $20 per member yearly ($12, $24 monthly) | Notion Agent on Business; Custom Agents "$10 per 1,000 monthly Notion credits" | Notion AI review |
| ClickUp | Full project management with AI as an add-on | Free; Unlimited $7 per user yearly; Brain AI +$9 per user yearly ($14 monthly) | 1,500 AI Super Credits per user a month on Brain AI; $10 per 10,000 more | No JAVIS review yet |
| Lindy | An AI teammate in Slack and email rather than a workspace | Free: $50 in credits for 7 days; Team $29.99 per active user a month | 3,000 credits per user a month, pooled; $10 per 1,000 more | Lindy review |
| Relevance AI | Sales and ops agent pipelines with approvals and evals | No free plan; Pro $19 yearly ($29 monthly) | Actions per tool run plus Vendor Credits at cost; BYOK on paid plans | Review coming |
The short version: Taskade is the cheapest way to put a whole small team on one AI workspace, and Genesis apps go further than Notion or ClickUp in turning a prompt into a working tool. Notion and ClickUp win on enterprise controls and compliance paperwork. Lindy and Relevance AI are better if what you actually want is agents doing outside work (email, Slack, sales research) rather than a place to organize work.
This review has no affiliate links. Taskade did not see it or pay for it.
There are no affiliate links in this article. Every Taskade link goes straight to taskade.com, its documentation or its YouTube channel, with no tracking parameters from JAVIS. The drawbacks, the refund wording and the cheaper or safer alternatives are written for the reader.
Hands-on status: JAVIS has not signed up for, built on or paid for Taskade for this review. Facts come from Taskade's pricing page, help center and docs, Terms of Service, Privacy Policy, Security page, changelog, blog and official YouTube channel; store figures from the US App Store and US Google Play; competitor prices from their own pricing pages; user sentiment from Trustpilot, Product Hunt and Hacker News, all as of Oct 4, 2026.
Start on Free with no card. If you upgrade, pay monthly first, because payments are final, and watch the Usage page.
Open taskade.comRead Taskade's security page and privacy policy, and ask support for the AI sub-processor list before you upload client data.
Read Taskade's security pageComparing AI workspaces and agent tools? These reviews cover the closest alternatives.
Each card opens a published JAVIS review.
Questions people actually ask before paying
How much does Taskade cost in October 2026?
Per taskade.com/pricing on Oct 4, 2026: Free is $0; Pro is $10 a month billed yearly ($120 charged up front) or $20 monthly; Business is $25 yearly or $50 monthly; Max is $100 yearly or $200 monthly; Enterprise is "Custom." Prices are per workspace, not per person.
How many people does a plan cover?
Free covers 2 users, Pro 10, and Business, Max and Enterprise unlimited users. Taskade says "AI usage is metered in credits, not seats," so everyone shares the plan's credits.
How many AI credits do I get, and what happens when they run out?
Free gets 6,000 one-time credits; Pro 50,000 a month; Business 150,000; Max 400,000. At zero, AI generation stops and you cannot overdraw, but published apps, projects and non-AI automations keep working. You can buy packs ("$5 /pack") or turn on Auto Top-Up.
Can I get a refund?
Generally no. The Sep 30, 2026 terms say "Payments are final," with exceptions only where the law requires a refund, for App Store or Google Play purchases (the store's process applies), or if Taskade ends your plan without cause. Billing errors can be raised within 30 days.
Is there a Taskade lifetime deal?
Not for new buyers. Past AppSumo and similar lifetime tiers keep unlimited workspaces, users, projects and storage, but Taskade says their AI credits were "a goodwill gesture" and not part of the original purchase.
Does Taskade train AI on my data, and where is it processed?
Taskade's terms say it "does not use your workspace content to train generalized AI models." Its privacy policy says the model you pick decides which provider processes a request, and that some providers are "based outside the United States, including in China." It does not offer customers a signed DPA.
Is Taskade SOC 2 compliant?
Not yet. Its security page says an independent SOC 2 audit "is not complete" and "We have no SOC 2 report to hand you yet." SAML SSO is available from Business; SCIM on Enterprise.
Does Taskade work with Claude or Cursor through MCP?
Yes, on every paid plan. The hosted server at taskade.com/mcp uses OAuth and works with Claude Desktop, Claude Code, Cursor, Windsurf and Zed. Each action an MCP client takes spends credits.
Where this came from, and when.
Pricing and billing: taskade.com/pricing with the Monthly and Yearly toggles each selected, the plan cards, the "Compare all features" table and every FAQ answer; taskade.com/learn/account/billing-faq (updated Sep 22, 2026); taskade.com/learn/account/free-plan; taskade.com/learn/genesis/model-credits (Sep 22, 2026); taskade.com/learn/agents/ai-usage; docs.taskade.com Credits & Billing; taskade.com/upgrade/credits; taskade.com/learn/account/cancel-subscription (Sep 2, 2026); help.taskade.com cancellation article (Mar 18, 2026). All checked Oct 4, 2026.
Policies: taskade.com/terms, taskade.com/privacy and taskade.com/security, each "Last updated September 30, 2026."
Product: taskade.com homepage, /about, /press, /mcp, /changelog (latest entries Sep 21 to Sep 25, 2026); blog posts "Introducing Taskade TSK-1" (Jul 8, 2026) and "How Taskade Became Taskade Genesis" (Sep 30, 2026); docs.taskade.com pages on Genesis, GitHub import and export, agent tools, AI Teams, Teams and roles, project views and GenesisAuth.
Images: Taskade's social image, press kit ("free to use when you write about Taskade"), Learn guides, homepage and documentation, plus Taskade's own US App Store screenshot. Product images use Taskade's sample data.
Videos: Taskade's official YouTube channel (@Taskade, linked from taskade.com): CCqBY4g8Ty4 "Build an AI App with Agents & Automations (No Code)" (Sep 2, 2026) and 8083rPm7vqM "Analytics, Custom Domains + 100+ Integrations After You Launch" (Sep 27, 2026).
Store ratings: Apple App Store, US storefront (app id 1264713923); Google Play, US store (com.taskade.mobile, gl=US, hl=en). Oct 4, 2026.
User sentiment: Trustpilot (trustpilot.com/review/taskade.com: TrustScore, distribution, labels, recent and low-star reviews); Product Hunt (producthunt.com/products/taskade/reviews); Hacker News comment search. G2, Capterra and Reddit not included.
Competitors: notion.com/pricing (both toggles), clickup.com/pricing (plans and AI pricing, both toggles), lindy.ai/pricing, relevanceai.com/docs/get-started/pricing, Oct 4, 2026.
JAVIS Fit Score (74/100): an editorial composite for a small team (3 to 15 people) wanting one AI workspace, weighing value for money (very strong: flat per-workspace pricing, real free plan), product breadth (strong: projects, mind maps, agents, automations, Genesis apps, MCP), credit transparency (mixed: hard ceiling and apps that keep running, but no public per-model price list and no Auto Top-Up cap), contract terms (weak: payments final, AI terms changed for past buyers), security and privacy (weak for business buyers: no SOC 2 report, no DPA, model routing that can include China-based providers) and support reputation (a deduction). It is judgment, not a measurement.
What this review does not cover: the quality, speed or reliability of Genesis builds in practice, the REST API in depth, App Kit marketplace economics for sellers, and Taskade's Enterprise contract terms.
