Taskade Genesis - Build live apps with agents, memory and workflow automation.

honest answer, not as a first-class versioning system yet

you can export app bundles as snapshots which gives you a manual versioning pattern. export before a big change, keep it as a backup

full version history with diffs and rollback is on the roadmap. its the right thing to build as apps get more complex and teams iterate faster

adding your vote for it internally. thanks for naming it specifically, this kind of feedback shapes what we prioritize

in the meantime start building at https://www.taskade.com/create

thanks for the congrats!

memory works at the Project level. each project is a structured database that your agents and automations read from and write to. its real persistent queryable data, not floating context in a chat window

cross-project: agents can reference multiple projects as their knowledge base. we shipped a dedicated Agent Knowledge tab in January where you configure what sources each agent pulls from. an agent in one app can access projects across your whole workspace

limits depend on your plan tier but for most use cases its effectively unlimited for structured data. and agent conversations are auto-saved across sessions now so no context is ever lost

build something and see it in action at https://www.taskade.com/create

marketing team is a great use case. content pipelines, campaign trackers, brief generators with automations that actually push the output somewhere useful

on transparency, we shipped AI Thinking Modes in December including a Reasoning Mode where the agent shows its full step-by-step logic before giving you the final answer. you can literally watch it think through the problem

we also added agent conversation export as Markdown in February so you can save and share any agent interaction for review

a more structured audit log for automation runs specifically is on the roadmap. thanks for naming it, it helps us prioritize

build something for your team at https://www.taskade.com/create and check out marketing tools others have built at https://www.taskade.com/community

thanks! and yes, this was important to get right

we have built-in rate limiting for automations with queue management to prevent hammering external APIs. each automation has retry logic and error handling baked in

for external API calls you can configure integration steps with timeouts and the system backs off automatically on rate limit responses. we also added If/Else branching and loops so you can build conditional retry logic at the workflow level

since this launch weve added 100+ integrations via the new Integrations Directory, MCP Connectors for even more extensibility, and custom agent tools so your agents can call external APIs as part of their reasoning chain

start building at https://www.taskade.com/create

do it! start with a real problem. a CRM, a client portal, a daily ops dashboard, whatever you actually need

one prompt at https://www.taskade.com/create and watch it assemble. you get a working app with agents, data, and automations built in

and if you want inspiration first check out https://www.taskade.com/community for apps others have already built. clone anything in one click and make it yours

not a silly question at all! its one of the most common ones we get

yes! we shipped file uploads in agent conversations in February. you can upload documents including XLSX, CSV, PDF, DOCX, PPTX, images and more. your agent reads the data and works with it directly in chat

for deeper spreadsheet workflows the Google Sheets integration is fully live and bi-directional. you can sync a sheet and have your agents and automations act on the data in real time

no developer skills needed for any of this. just describe what you want at https://www.taskade.com/create and go from there

2017! youve been here through all of it and that genuinely means a lot

on more detailed explanations, totally fair ask. a few things weve shipped since then:

AI Thinking Modes launched in December with four modes including a Reasoning Mode where the agent walks through its full step-by-step logic before giving you the answer. really helps with understanding whats happening under the hood

Agent Knowledge tab shipped in January so you can see exactly what sources your agent is pulling from and manage its training data

Agent conversation export as Markdown shipped in February so you can save and review any interaction

would love to hear what specific areas still feel opaque. your perspective as someone whos been here this long is really valuable

check out whats new at https://www.taskade.com/community or build something at https://www.taskade.com/create

really good questions, both partially live and actively being built out

portability: app bundle export/import is live. you can package the whole app structure and move it across workspaces. full API access for workspace data is where were going next. not fully standalone yet but the data is always yours

guardrails: per-model credit tracking is live so you can see exactly what each agent costs. custom agent tools launched in February which lets you tightly scope what each agent can actually do. no more agents with blanket access to everything

we also shipped 12 AI models you can choose from (speed-optimized, reasoning-optimized, writing-optimized) so you can match the model to the task and control spend that way

cost caps per agent specifically are on the roadmap

try it at https://www.taskade.com/create and let me know if anything else is blocking you