Changelog

Follow new updates and improvements to Swiftask.

March 4th, 2026

We've been listening. You told us meetings needed to feel more alive. This month, we made your recording sessions smarter, faster, and more interactive.

Enhanced meeting recording interface

We've improved the meeting recording experience with a redesigned interface and real-time collaboration features. You can now have discussions with AI while your meeting is actively recording, and see your transcription appear progressively as you speak.

  • Improved UI element positioning and design in the meeting interface

  • Real-time discussion capability during active meetings

  • Live transcription that updates every minute during recording

  • Enhanced chat panel for asking questions about meeting content while recording

Available on all plans when you start a meeting in Meetings โ†’ Record Meeting.

How to test it:

  1. Navigate to Meetings in the left sidebar

  2. Click "Record Meeting" to start a new recording

  3. Notice the improved interface layout and design

  4. Speak naturally and watch the transcription appear progressively

  5. Use the chat panel on the right to ask questions about your meeting content in real-time

  6. Experience the enhanced real-time discussion features during your active recording

Improvements

  • Meeting interface elements have been repositioned for better usability and visual clarity during recording sessions

  • Transcription now appears progressively during meetings rather than only after recording ends, updating approximately every minute for immediate feedback

  • Real-time AI discussion capabilities allow you to interact with meeting content while recording is in progress, making meetings more interactive and productive

March 4th, 2026

Good news for your AI projects: you no longer have to choose between maximum power and cost control.

Three new Claude models are now available via AWS Bedrock in the Europe region. Each addresses a specific need: cutting-edge intelligence for your strategic projects, economic efficiency for high-volume tasks, or the perfect balance between the two.


๐Ÿš€ Claude 4.6 Sonnet โ€” Cutting-edge intelligence made accessible

Performance nearly equivalent to Opus (our most advanced model) at the Sonnet 4.5 price: 3/3/3/15 per million tokens.

Ideal for:

  • Developing complex autonomous agents

  • Code generation with complete architecture

  • Strategic analysis of large documents (up to 1M tokens)

Concrete result: Reduce time spent on refactoring and code migrations by 40% thanks to its multi-step reasoning capability.


๐Ÿ’ฐ Claude Haiku 4.5 โ€” Economic efficiency without compromise

Performance equivalent to Sonnet 4 (our former premium model) at 1/1/1/5 per million tokens โ€” 3ร— cheaper than Sonnet models.

Ideal for:

  • Large-scale automated customer support

  • Real-time document summaries

  • Intelligent monitoring and alerts

Concrete result: Divide your AI infrastructure costs by 3 on recurring tasks, while maintaining professional quality. At 100,000 sessions/month, save $4,500 compared to Sonnet 4.5.


โš–๏ธ Claude Sonnet 4.5 โ€” The optimal balance

The best compromise between intelligence and cost, for most of your daily workflows.


Capabilities common to all three models:

  • Deep reasoning (extended thinking) for complex tasks

  • Web interface automation (computer use)

  • Image and visual document analysis

  • Available on all Swiftask plans


Improvements

Enhanced GDPR compliance Hosting on AWS European servers (EU region) for optimal latency and full compliance with European data protection regulations.

Smart budget optimization With Haiku 4.5, reduce your operational costs by up to 66% on high-frequency workflows (customer service, stream analysis, automated monitoring) without sacrificing response quality.


How to test: Open any chat or agent โ†’ Click the model selector โ†’ Choose from the new Claude Europe models. All your existing skills, files, and automations work immediately.

Our advice: Start with Haiku 4.5 for your recurring tasks, and reserve Sonnet 4.6 for projects requiring advanced reasoning. This way, you'll optimize your budget while keeping power where it matters.


March 4th, 2026

We added a "Duplicate" option to all slide artifacts. You can now create presentation templates once and reuse them across your entire teamโ€”no more rebuilding decks from scratch.

How it works:

  • Duplicate any slide instantly

  • No limits on the number of copies

  • Duplicates appear in your library

  • Share templates with your teammates

Use cases:

  • Sales pitch: One unified pitch template for your entire sales team

  • Recurring reports: Same structure, only the data changes

  • Onboarding: New hires start from your best presentations

Where to find it: Artifacts โ†’ select a slide โ†’ โ‹ฎ โ†’ Duplicate

Try it now: Open a slide, click the three-dot menu โ‹ฎ, select "Duplicate". Your template is ready.

February 27th, 2026

Building effective agent prompts just got easier. We've added an AI-powered prompt optimizer that analyzes your agent instructions and suggests specific improvements, complete with a ready-to-use optimized version you can apply instantly.

AI-powered prompt optimization for agents

You can now get instant, actionable recommendations to improve your agent prompts with the new "Check & optimize" button. An AI agent analyzes your prompt and provides specific suggestions alongside a fully rewritten optimized version, making it simple to build more reliable and effective agents.

Split-view optimization interface

When you click "Check & optimize," a modal opens showing:

  • Recommendations panel โ€” Bulleted list of specific improvements (remove formatting artifacts, complete missing sections, generalize hardcoded data, add structured messages, specify reference URLs)

  • Optimized prompt panel โ€” Complete rewritten version incorporating all recommendations

  • Apply optimized prompt button โ€” One-click to accept all changes, or close to keep your original

Version history

Save your prompt iterations using the "Name this version" button to maintain a complete history of your prompt evolution and easily roll back if needed.

Available on all plans with no usage limits.

Find it in Agents โ†’ [Agent Name] โ†’ Edit โ†’ Instructions, in the top-right area next to "Name this version."

How to test it: Open any agent, go to Edit โ†’ Instructions, click "Check & optimize" in the top-right corner, review the AI-generated recommendations and optimized prompt side-by-side, then click "Apply optimized prompt" to accept the improvements or close the modal to keep your original.

February 18th, 2026

Planning your agent automations just got a whole lot easier. We've added a visual calendar interface that transforms how you schedule and manage your AI workflows, making it simple to see your automation timeline at a glance.

Calendar view for agent automation

You can now visualize and schedule your agent automations using a clean calendar interface instead of the traditional table format. Perfect for planning recurring tasks, scheduling one-time automations, and getting a visual overview of your automation timeline.

  • Switch between table and calendar views with toggle buttons

  • Click any date to create an automation with pre-filled start date

  • Navigate by month, week, or day with dedicated view controls

  • Filter automations to find specific scheduled tasks

  • Visual indicators show existing automations on calendar dates

  • Navigate between months using arrow controls or jump to today

Available on all plans with no limits on the number of automations displayed.

Find it in Agents โ†’ [Agent Name] โ†’ Automations โ†’ Calendar view icon.

How to test it: Open any agent, go to Automations, click the calendar view icon (next to the table view icon), then click on any date to create a new automation and see how the start date gets automatically filled.

Improvements

Enhanced automation scheduling workflow with visual calendar interface makes it easier to plan and organize your agent tasks across time, helping you avoid scheduling conflicts and better manage automation timelines.

February 18th, 2026

We've added a new way to keep your agent costs predictable and under control. Whether you're running customer-facing widgets or public agent apps, you can now set daily credit budgets to prevent surprises on your bill.

Daily credit limits for agents

You can now set daily credit limits for individual agents to prevent unexpected credit consumption and control costs. This works for both widget deployments and public agent apps, helping you maintain budget control while keeping your agents accessible.

When an agent reaches its daily limit, users see a clear message indicating the limit has been reached and when they can interact again.

  • Set any credit amount you want โ€” you choose minimum or maximum restrictions

  • Applies to both widget and agent-as-app usage

  • Limits reset daily at midnight

  • Available on all plans

Find it in your agent editor โ†’ More options โ†’ Cost control

How to test it: Edit any agent, navigate to More options โ†’ Cost control, enable "Daily budget limit," set a low test value (like 100 credits), then interact with the agent via widget or public link until you hit the limit to see the restriction message.

February 18th, 2026

You can plan your projects in the Projects section on Swiftask and organize your tasks in Kanban mode with AI assistance. We've now significantly enhanced task creation and agent assignment. Discover how to get the most out of it.

Tell your agent to create a task in your Kanban

Create tasks by voice-dictating your request directly to an AI agent. State your mission in one sentence and the agent automatically generates the task title and description. This new conversational interface replaces the old multi-step form with a unified, intuitive flow.

Real example: Say "Create a client presentation" and the agent instantly generates the complete task with title and description.

Features:

  • Automatic voice transcription with no time limits

  • Support for all languages

  • Integrated agent selection in the creation process

  • AI-powered automatic title and description generation

  • Instant delegation to your selected agent

To try it out:

  1. Go to Projects โ†’ [Your project] โ†’ + Add task

  2. Click on agent selection in the bottom left

  3. Enable voice input (microphone icon)

  4. Dictate your task in one sentence

  5. Send with the arrow button (bottom right)

The agent will automatically create the complete task and be immediately available to assist you.


Improvements

Unified creation interface

No more complex forms with separate fields for title, description, and assignment. Now create your tasks in a chat-style interface that combines all these steps into a single conversational flow. Result: faster task creation with less friction.

Streamlined team assignment

Assign tasks to your team members directly through the task setup. Each task has a dedicated chat where agents can automatically start assisting from the moment of creation.


Availability

All users โ€” This feature is available to all Swiftask users, regardless of their plan.

All platforms โ€” Works seamlessly in browser and desktop app.

No prerequisites โ€” No setup required.


Ready to try it? Open a project right now and click + Add task to discover the new voice interface. Your AI agents are waiting for your instructions.

February 17th, 2026

Your email writing just got smarter. We've enhanced the Swiftask Chrome extension with instant AI-powered improvements that work directly in Gmail and Outlook, making it even easier to craft perfect messages while you compose.

Three powerful quick actions now at your fingertips:

  • Improve writing โ€” Enhances clarity, tone, and overall message effectiveness

  • Fix spelling & grammar โ€” Corrects errors and improves language accuracy

  • AI reply โ€” Generates contextual responses based on the email thread

Quick actions are available in three convenient ways: buttons at the top and bottom of your text area, plus right-click anywhere in your compose window. This builds on your existing agent-powered email drafting capabilities, giving you even more control over your communication.

How to test it:

  1. Open Gmail or Outlook in Chrome

  2. Start composing or replying to an email

  3. Look for the quick action buttons above and below your text area

  4. Select any quick action to see your text improve in real time

Available with the Swiftask Chrome extension on all plans.

Improvements

Faster agent email processing โ€” Agent email addresses now process messages faster with enhanced parsing of complex email threads and attachments, for better context understanding and more accurate automated responses.

February 5th, 2026

Here's everything that happened recently with Swiftask that you should try this month:

Artifact public sharing

As a reminder, we created the Artifacts feature found in the left sidebar (under "Knowledge") to easily create and find all your PowerPoint presentations and websites.

Now, you can share your presentations and web pages publicly with anyone, even without a Swiftask account. Perfect for sharing client presentations, marketing materials, or project showcases with external stakeholders.

  • Share both slide presentations and interactive web pages

  • Recipients access content via direct link without login required

  • Public links work across all devices and browsers

  • Available on all plans

Find this feature in Artifacts โ†’ Open your slide or web page โ†’ Share โ†’ Public.

To test: Share an artifact link with a colleague outside your workspaceโ€”they can open and view it immediately without needing a Swiftask account.

Useful improvements have also been made

Enhanced email automation with CC and BCC support

We all know that we can attach the SMTP skill to an AI agent you create so it can automatically send and reply to emails.

Using the improved SMTP skill, agents can now send emails with CC and BCC recipients. This enables more advanced email workflows like keeping managers in the loop or sending blind copies to compliance teams.

  • Add multiple CC and BCC email addresses

  • All email addresses go through validation to ensure they're properly formatted

  • All messages are identical for all recipients

  • Works with any SMTP-compatible email provider

To use the SMTP functionality, go to Agent modification โ†’ Skills โ†’ Add skills โ†’ Communication โ†’ SMTP.

To test: Ask your agent to send an email and include CC/BCC recipients to verify the functionality works correctly.

Your agents can now trigger automatic actions

When creating your agents, you can configure an extended trigger system so they act automatically at the slightest detected movement. These triggers include Outlook email monitoring, webhook integrations...

New trigger types:

  • Webhook โ€“ Trigger via HTTP webhook URL for integration with external services

  • Email (IMAP) โ€“ Trigger when new emails arrive in your inbox. Compatible with Gmail and other IMAP providers

  • Outlook 365 Email โ€“ Monitor your Outlook inbox using modern OAuth2 authentication

  • App Integration โ€“ Connect to 2000+ apps including Slack, GitHub, and Google Sheets

Configure triggers in Agents โ†’ [Agent Name] โ†’ Edit โ†’ Triggers.

To test: Set up an Outlook trigger and test it by having a friend send an email to the monitored Outlook address.

Add multiple Google accounts to Google Drive exploitation skills

As a reminder, you can attach skills to your agents that require Google account integration such as: automatic Google Sheet creation, creating files or folders in Google Drive...

Now, you can add multiple Google accounts so you can choose which Drive the agent will work in.

To make it easier for you to manage your Google accounts integrated with Swiftask agents:

  • Connect multiple Google Drive accounts to a skill that requires it

  • Agents can access files from each of the connected accounts

  • Improved account management

To test: Add a Google Sheet creation skill to any agent, add multiple Google Drive accounts, and choose which of these accounts it will create the file in.


All features are available on Pro, Team Starter, Team Growth, and Custom plans.

January 31st, 2026

Each AI agent can now receive and process emails directly through a unique email address. This feature transforms agents into intelligent email assistants that can automatically handle incoming messages without manual intervention.

Key capabilities:

  • Unique email address per agent: Every agent gets a dedicated email address (e.g., agent-slug@swiftask.ai) allowing direct communication via standard email clients

  • Automatic email processing: Agents automatically read, parse, and extract information from incoming emails including sender details, subject lines, message content, and attachments

  • Intelligent response generation: Agents can analyze email context and automatically generate appropriate responses based on their configured knowledge base and instructions

  • Email thread management: Handles email conversations and maintains context across multiple exchanges

  • Attachment support: Process documents, images, and files sent as email attachments

Use cases:

  • Document processing: process orders, invoices from email automatically

  • Redirect any tasks from email to an ai agent

  • Lead capture automation: Forward inquiry emails to your sales agent for automatic lead qualification and CRM updates

  • Customer support triage: Route support emails to specialized agents that categorize, prioritize, and draft responses

  • Content collection: Send marketing materials, competitor emails, or campaign reports directly to your research agent for analysis

  • Event registration processing: Forward event signup emails for automatic attendee management and follow-up sequences

  • Newsletter management: Have agents process subscription requests, feedback, and engagement data from email campaigns

How it works:

  1. Navigate to your agent settings in the Agent admin interface

  2. Copy your agent's unique email address from โ€˜Triggersโ€™ menu

  3. Forward emails or set up automatic email forwarding rules (Gmail, Outlook, etc.)

  4. The agent receives, processes, and responds according to its configuration

  5. Usage history is visible from the โ€˜Chatโ€™ of the agent owner

Technical Details:

  • Supports all standard email formats (plain text, HTML, multipart)

  • Multi-language email processing capability

  • Integration with existing agent skills, knowledge bases, and automation triggers

Location: Available in Agent admin โ†’ Profil or Agent / Triggers

This feature is available for all plans