Tasklet

Run your business on AI agents. Connect your tools, automate work 24x7, and share agents across your whole company.

[Tasklet @TaskletAI]

Sunday: CEO tries Tasklet.
Monday: Emergency Onboarding.

30 mins later, each employee had built a first agent.
40 days later, 60+ people actively using agents.

Here’s the story ↓

[Mark Robinson @markallenrob]

how i order lunch these days:
"pick something tasty that's not super messy"

Our office Tasklet agent sends the group order in Slack and handles the rest 🍲

[Andrew Lee @startupandrew]

Tasklet is a model-agnostic harness that gives you the best value per dollar across Anthropic / OpenAI / Google / xAI / etc.

Now we automatically route tasks to the most effective model for you.

This includes cross-model delegation for subagents. For example, the auto-router might pick Claude Opus 5.5 for the main agent but delegate work to GLM 5.3, GPT 6.1 Sol, and Gemini 4 Argon under the hood where it makes sense.

[Albert Wenger 🌎🔥⌛ @albertwenger]

Love not having to worry about whether I am using the right model for the task!

[Tasklet @TaskletAI]

4/

We keep tuning Auto mode as new models launch, so you can focus on getting work done instead of tracking who's winning the AI race this week.

New agents use Auto mode by default, and you can still pick a specific model anytime.

See how it works →
https://t.co/oc3En3eO7C

[Tasklet @TaskletAI]

1/

Introducing: Auto mode

Tasklet now picks the best model for every task, automatically. Each message gets handled by the model that can do it well, for the lowest price.

And with Ultrafast, you can turn up the speed when the work can't wait.

[Tasklet @TaskletAI]

3/

Moving quick? Go Ultrafast.

Auto mode picks the model, you pick the speed.

Choose Ultrafast when you want to stay in flow and the work can't wait. It uses about 6x the credits, so Auto mode never turns it on for you. You decide when speed is worth it.

[Tasklet @TaskletAI]

Our integration with @mercury is now live.

Now you can get daily financial reports, analyze spending & burn, reconcile payments, flag unusual transactions, and more, right from Tasklet.

[Tasklet @TaskletAI]

Saved logins are encrypted and only filled on the matching site. Within a shared agent, everyone using it can use them, so only add the business accounts you're comfortable sharing.

Save the login once. Let your agent take it from there.

Read the blog:
https://t.co/bzxNrTY7aL

[Tasklet @TaskletAI]

Your agent used to get halfway through a browser task, hit a login screen, and wait for you.

Not anymore. Securely save a login once in Tasklet, your agent can sign itself back in and keep going.

Three features we added to fix this: 🧵
https://t.co/bzxNrTY7aL

[Tasklet @TaskletAI]

3. Give your agent a 1Password vault

Instead of saving logins one at a time, give your agent a dedicated vault with the credentials it needs.

You keep managing them in @1Password and anything you add later is available to your agent automatically.

[Tasklet @TaskletAI]

Gemini 4 Argon is live in Tasklet.

Google’s new frontier model is built for long, multi-step work across finance, legal, and video.

Your agents shouldn’t belong to one model. Select Argon on any agent—your connections, skills, and context stay put.

[Google @Google]

Today, Gemini 4 Argon is officially rolling out.

[Tasklet @TaskletAI]

@elonmusk @bot @SpaceX

[Tasklet @TaskletAI]

Claude Haiku 5.5 is live in Tasklet ⚡️

Your agents can use the best model for each task, across labs. Lower-cost models for routine work. More capable models for the hard stuff.

[Claude @claudeai]

Introducing Claude Haiku 5.5: the cheapest, fastest, and most capable small model we’ve ever released.

On average, it costs around 75% less to run than Claude Haiku 4.5.

[Colt @ColtSauers]

@_renatov @thogge For anything requiring integrations to other tools (email, notion, etc) @TaskletAI is A+. It’s unreal how well it just *works*

[Gaurav Jain @gjain]

@startupandrew @TaskletAI I am addicted to @TaskletAI!!

[Rakesh Agrawal @RakeshAgrawal]

We’ve got 100+ people building agents at @pipingtech with @TaskletAI. I’d be using it whether or not I had invested. The product was amazing when I started using it in May and the Tasklet team’s pace of development is phenomenal. No better way for normal people and teams at companies to build AI agents than Tasklet. Not that I’ve seen anyways!

[Tasklet @TaskletAI]

A few things we recently shipped!

Your Tasklet agents can delegate to multiple subagents in parallel while you keep working with it.

Also new: a refined UI, sidebar with more visibility into your work, and public links to share artifacts and files.

Here's a walkthrough 👇