Tasklet's new sandbox VM: process files, analyze data, generate media & more

With today’s launch, every Tasklet agent now has its own filesystem and VM sandbox—a real Linux environment where it can run terminal commands and execute code.

Most AI assistants generate code and hand it back to you. Tasklet writes the code, runs it, catches errors, debugs, and retries—automatically. Ask for a chart and get a chart. Ask for a report and get a downloadable file. The gap between “here’s how you could do it” and “here, I did it” just closed.

Files now flow between your connections

We’ve upgraded connections to support files. You can now move files between Gmail, Google Drive, Slack, APIs, computer use, and the agent’s filesystem seamlessly.

Combined with the new sandbox, this unlocks a whole category of file workflows:

• Transfer — Grab email attachments and file them to Drive, or download files from the web and upload to Slack
• Convert — Batch resize images, transcode audio, turn a folder of documents into PDFs
• Create — Generate an Excel spreadsheet, a PDF report, or a video with narration from scratch
• Edit — Modify images, merge PDFs, clean up data files
• Analyze — Process a 10,000-row CSV and get back summary statistics and visualizations

All of this happens inside Tasklet—no manual downloads, no switching between apps.

Automate it with triggers

The sandbox gets even more powerful when you combine it with triggers. Set up a workflow once and let it run:

• Every morning, pull data from your CRM and generate a summary report
• When receipts arrive in your inbox, extract the data and add a row to a spreadsheet
• On a schedule, check a public dataset and alert you if anything changes

Complex data pipelines that used to require engineering time now run autonomously.

Computer use + sandbox = magic

Tasklet’s computer use lets your agent control a real browser. Now it can take what it sees and do something with it:

• Grab screenshots of a competitor’s website and generate a design critique as a PDF
• Navigate to a dashboard, export the data, and run custom analysis
• Record a workflow, then produce a video walkthrough with narration

The sandbox turns computer use from “look at this” into “look at this, now process it.”

What’s under the hood

Every agent now has access to:

• A Linux sandbox (Alpine) with Python 3.12, shell, and full network access
• Preinstalled CLI tools — ffmpeg for audio/video, imagemagick for images, pandoc for document conversion, poppler for PDFs
• Persistent storage — files survive across sessions
• Any package you need — the agent can install additional tools on the fly

Available now for all Tasklet users, on every plan.

Have feedback? Drop me a note at jonny@tasklet.ai.