Tasklet's new v2 agents: smarter, more capable, and easier to use
by Andrew Lee
When we launched Tasklet two months ago, we set out to build AI agents that actually do the work—not just answer questions, but connect to your tools, run autonomously, and handle the complexity for you.
Since launch, we’ve heard a lot from you. You want agents that can handle more triggers, tackle more complex projects, work around errors more robustly, and get more done autonomously.
Today, we’re releasing our v2 agent architecture: a ground-up rebuild of how agents think, plan, and execute work.
What’s new
Multiple triggers per agent
Previously, each agent could only configure one trigger at a time. Now a single agent can handle unlimited triggers—run a daily briefing every morning, respond to incoming emails, and react to webhooks from your CRM, all in one place.
No more duplicating agents to handle related workflows.
Agents that plan their own work
The v2 agent has a built-in task list that lets it break down complex projects, queue up work, and track progress across multiple steps. When a trigger fires, it creates a task. When you ask for something complicated, the agent can decompose it into manageable pieces and work through them systematically.
This helps agents handle multi-step projects that would have overwhelmed the previous version—think “research 50 companies and summarize findings” or “process this week’s support tickets and categorize them.”
Subagents for complex workflows
Agents can now spin up specialized subagents on the fly to tackle different parts of a problem. Need to process a batch of documents? Your agent can delegate each one to a subagent, then synthesize the results. This keeps agents fast and focused, even when the work gets complicated.
Smarter context management
We completely rebuilt how Tasklet manages your working context, enabling agents to process much larger amounts of data, use quota more efficiently, and to keep running smoothly even for long, complex tasks.
A simpler experience
All of this power comes with a cleaner interface. Trigger executions now appear inline in your conversation instead of hiding in a sidebar. You can even queue up multiple messages without stopping the agent, and the stop button actually stops immediately.
But wait, there’s more…
We’ve been busy. Here’s what else we’ve shipped since launch:
- More integrations — an expanded library of Tasklet-built connectors, so more of your tools work out of the box
- Better MCP discovery — official MCP servers from providers like ClickUp, Linear, and others are now easier to find and connect
- More API auth options — OAuth, API keys, basic auth—connecting to custom APIs just got more flexible
- Broader file support — Your agent can now work more effectively with PDFs, markdown files, images, and more
- Performance & reliability — dozens of bug fixes and under-the-hood improvements that make everything feel snappier
How to get the new agent
New agents you create today automatically use the v2 agent architecture. Try it right now. For existing agents, you’ll see a banner at the bottom of your conversation—just click “Update agent” and you’ll be ready to go in moments.
Have feedback? Drop me a note at andrew@tasklet.ai.