Turn a plain-language description into a hosted business app with connected data, AI agents, and automated workflows. Taskade Genesis brings app building into a collaborative workspace.
Used this tool? Rate it
Used this tool? Rate it
Taskade is a workspace for building business apps, organizing their data, and connecting AI work to business processes. Taskade Genesis is the AI app builder inside Taskade, turning a plain-language description into a hosted app with data, AI agents, automations, and an interface. Its current scope includes CRM systems, client portals, dashboards, and booking apps alongside the projects and task views used to manage the underlying work.
Workspace DNA connects projects as Memory, AI agents as Intelligence, and automations as Execution; the app interface supplies pages, forms, and dashboards. The distinction matters when describing an app: a contact record belongs in a project, an agent can score that contact, and an automation can act when the record changes. The visible app brings those parts together for its users.
Projects support List, Board, Table, Mind Map, Docs (Beta), and Calendar on every plan; Gantt Chart, Actionsheet, and Org Chart require a paid plan.
Custom agents combine instructions, a model, knowledge sources, saved prompt commands, and tools such as search, project actions, apps, and automations. Instructions define the agent's role and response style, while commands provide saved prompts that can be invoked with slash shortcuts.
Agent knowledge can come from projects, files, links, videos, and pasted text.
Automations connect a trigger to actions and outputs, with filters, branches, loops, and delays; actions can update projects, call AI, or use integrations.
Automation triggers include workspace events, schedules, inbound email, webhooks, forms, AI agent chats, and connected apps.
The Flow tab is for assembling an automation, while Runs records the data flowing through each step and offers one-click retry.
App secrets are available on every plan; reading, copying, and editing full app source in the Code tab requires Pro or above. Secrets provide a way to call an outside service without putting its key into app code.
A Genesis build starts with users, data, actions, and a result, creates projects, agents, automations, and pages together, and can be tested in the live preview. The documented build sequence keeps the app's behavior tied to its data and intended audience:
Taskade describes a CRM combining contacts and deals in projects, AI lead scoring, automated follow-ups across Gmail, Slack, and Calendly, and a published dashboard. The CRM example ties recordkeeping to agent judgment and external communication: the contacts table holds the pipeline, the agent supplies a score or draft, and the automation carries the next action.
A documented booking-app example collects a service, date, and time slot, stores bookings in a project, and sends a Slack message when a booking arrives. It demonstrates a narrower process than a whole CRM: a customer-facing input, a persistent booking record, and a notification to the team.
Taskade Genesis is positioned for operators without coding experience, including a CRM that tracks leads and follows up automatically. Its natural-language starting point fits teams that can describe their process in terms of records, users, and actions, even when they do not want to begin by writing the application source.
These collaboration roles concern access to the building workspace and its content. A customer using a published app has a separate access path, and visitor sign-in has its own data-scope controls. That distinction matters for a client portal as much as for an internal tool: a colleague's editing role and a client's record visibility answer different questions.
Taskade offers desktop apps for macOS, Windows, and Linux, mobile apps for Android and iOS, and browser extensions for Chrome, Firefox, and Edge.
GitHub import expects an exported Taskade app repository, not an ordinary code repository, and requires connecting a GitHub account even for a public repository.
The following public prices were captured on October 28, 2026. Annual prices are monthly equivalents charged as a full year; the monthly column is a separate, explicitly listed billing option.
| Plan | Monthly equivalent, billed annually | Annual charge | Billed monthly |
|---|---|---|---|
| Free | $0 | $0 | $0 |
| Pro | $10 | $120 | $20 |
| Business | $25 | $300 | $50 |
| Max | $100 | $1,200 | $200 |
| Enterprise | $250 | $3,000 | $500 |
Free starts with no AI credits; paid monthly allowances are 50,000 on Pro, 150,000 on Business, 400,000 on Max, and 1.1 million on Enterprise, with members bundled up to each plan limit.
Free includes 3 apps, 2 people, and 10 monthly automation runs; Pro adds unlimited apps, 10 people, 50,000 monthly AI credits, and 1,000 runs; Business adds unlimited people and 10,000 runs, while higher tiers increase AI capacity. Pro includes up to 10 custom domains, and Business adds unlimited custom domains, app and agent branding, and SAML single sign-on. Max lists four times Business’s AI request limit and auto top-up of up to $1,000 at a time; Enterprise adds SCIM provisioning. The request limit is distinct from the monthly credit allocation.
Free allows one custom AI agent; Pro and above allow unlimited agents, AI Teams, and incoming webhook triggers.
Free includes HTTP Request, RSS, Pushover, and MCP Client connectors; premium connectors such as Gmail, Slack, and Stripe run on paid plans. The MCP Client connector is distinct from access to the hosted MCP server.
TSK-Auto and TSK-Instant are available on every plan; TSK-Standard and TSK-Thinking require a paid plan, and TSK-Reasoning requires Business or above. Model selection changes the credit cost and reasoning depth, so an allowance is not a fixed number of interchangeable requests.
Monthly Credits are used before Extra Credits and reset each billing cycle; purchased or earned Extra Credits do not expire. Monthly Credits reset on the billing anchor date; when credits run low, non-AI features keep working and only AI actions pause until credits reset or are added.
Taskade says published Genesis apps run free for visitors, while building and AI work use credits and the model picker shows the current cost. AI-powered automation actions consume workspace credits, while non-AI workspace actions, integrations, and control-flow steps do not. The free-running published-app statement therefore sits alongside a separate rule for AI actions, rather than establishing unlimited AI execution.
Cancellation in Settings → Billing stops future charges, keeps the paid account active through the paid period, and then returns it to Free; app-store subscriptions are handled through their store. Cancelling does not retroactively refund an existing charge.
Payments are generally non-refundable, including subscriptions, renewals, AI credits, unused seats, and paid App Kits, with exceptions for applicable law, app-store purchases, and termination of a paid plan without cause. The last exception provides a pro-rata refund of the unused paid-plan portion. These are the terms' stated exceptions, rather than a general refund window.
The alternatives address two different starting points: organizing knowledge and developing an application with editable code.
Public apps are accessible to anyone, Secret apps are hidden from search and galleries but accessible by link, and Private apps stay inside the workspace.
Public visibility does not make an app cloneable; Let others clone is a separate setting with Free or Paid options.
Visitor sign-in establishes identity but does not by itself hide other people’s records; row scoping and matching records to the login are separate controls. A login screen alone therefore does not establish that each client can see only their own records. The published app's identity and data-access configuration determine different parts of that boundary.
Taskade describes automatic backoff retries for transient automation failures, but an agent step runs once with up to 15 minutes to finish.
Five consecutive failed runs turn an automation off and email workspace owners, with earlier notices after the first and third failure; documented exclusions include filtered runs, Taskade rate limits, runs not started due to plan limits or being off, and failures in Taskade’s built-in search service, while outside-app rate-limit errors still count. A successful run resets the consecutive-failure count.
Taskade says its independent SOC 2 audit is not complete, no SOC 2 report is available, and it has no fixed external audit cadence. This is a statement of the current audit position, not an independent certification.
Taskade does not currently offer a signed customer DPA or Standard Contractual Clauses, cryptographically immutable logs, tamper detection, or prebuilt compliance report formats. Taskade describes itself as a general-purpose product that is not built or certified for regulated data and does not scan for or automatically remove it.
Taskade states that it does not use workspace content to train generalized AI models and that its provider agreements prohibit training general models on that content; it may use de-identified or aggregated usage data to operate, secure, and improve the service.
The selected model determines which provider processes a request, and some AI providers are outside the United States, including in China; contractual training restrictions do not imply local-only processing. The privacy policy says AI providers follow their own policies, which may change, and disclaims liability for practices beyond Taskade’s contractual control. These are vendor commitments and stated limits, rather than independently verified provider behavior.
Taskade aims to remove account information within 30 days after account deletion; this is a target rather than a fixed contractual period, with backup, legal, dispute, investigation, technical, and shared-data exceptions. AI interaction logs are normally kept for no more than 30 days, but encrypted backups and security investigations or legal obligations can extend retention.
As between the user and Taskade, AI outputs belong to the user subject to third-party rights and applicable law; the terms require review for accuracy and suitability and warn that outputs can be inaccurate, incomplete, biased, or inappropriate. The service does not provide professional legal, medical, or financial advice. Ownership is not presented as a guarantee that generated content is accurate or suitable for a particular use.
Users retain ownership of their content, but public content gives Taskade a broad perpetual license; the separate user-to-user license covers public community surfaces and excludes paid kits or paywalled content, and Taskade’s license does not authorize giving a priced kit away free. The public-content license includes use, modification, distribution, derivatives, display, and promotion. Publishing community material thus has a different licensing scope from keeping it within a private workspace.
For data submitted to someone else’s published Taskade app, the app owner is the data controller, its privacy notice governs, and Taskade processes the data on its behalf.
App version history restores earlier design and code while leaving data as it is; it is available on every plan and editors can restore.
Taskade says disconnecting an integration stops data sending and removes stored credentials; supported integrations use OAuth scopes where offered, otherwise API keys or signed webhooks.
Taskade EVE remembers the selected model for the next conversation and falls back to TSK-Auto if the saved model is no longer available.
