Skip to main content

🧪 Beak AI Studio – Release Notes v1.1.0

Introduction​

Beak AI Studio v1.1.0 is a landmark release, introducing a suite of advanced features and architectural enhancements that empower users to build, orchestrate, and monitor sophisticated AI solutions. This version focuses on extensibility, agent autonomy, robust security, and seamless integration with enterprise data and workflows. Below, you’ll find comprehensive documentation of all major new features, with practical examples and guidance for both end users and administrators.

🚀 Major Features & Enhancements​

1. MCP Server Support​

Overview:
Beak AI Studio now allows users to add, configure, and manage Model Context Protocol (MCP) servers directly within the platform. MCP servers provide specialized skills and tools that can be assigned to agents, enabling seamless orchestration across multiple endpoints.

Key Capabilities:

  • Register and manage multiple MCP servers from the UI.
  • Assign specific skills from MCP servers to agents or agent collectives.
  • Dynamically route agent requests to the appropriate server/skill based on task requirements.

Example:
Configure a custom MCP server for document summarization, and assign it to a research agent for on-demand use.

Documentation:
See Skills: MCP Servers for setup and integration details.

2. Flexible Uploaded File Datasource​

Overview:
Users can now choose whether to leverage a vector database for uploaded files, or proceed without one. This flexibility allows for tailored storage, indexing, and search strategies based on project needs.

Key Capabilities:

  • Toggle vector database usage per file datasource.
  • Use traditional file storage for simple retrieval, or enable vector search for semantic queries.

Example:
Upload a set of PDFs and select “No Vector DB” for simple keyword search, or enable vectorization for advanced semantic retrieval.

Documentation:
See Data Sources: Files.

3. Expanded Web Datasource Capabilities​

Overview:
Beak AI Studio now supports ingesting content from public websites, web portals, and SharePoint sites as datasources. Both structured and unstructured web content can be integrated into AI workflows.

Key Capabilities:

  • Add URLs or SharePoint site links as datasources.
  • Automatically crawl and index web content for agent use.
  • Support for authentication and custom crawl settings.

Example:
Integrate your company’s SharePoint knowledge base as a datasource for an internal support agent.

Documentation:
See Data Sources: Websites.

4. Multi-Agent and Architecture Support​

Overview:
Comprehensive support for multi-agent compositions, including orchestration and swarm architectures. Design and manage agent collectives capable of parallel task execution and complex cooperative behaviors.

Key Capabilities:

  • Create agent groups with defined roles and communication protocols.
  • Orchestrate tasks across multiple agents for efficiency and redundancy.
  • Support for both centralized (orchestrator) and decentralized (swarm) models.

Example:
Deploy a swarm of research agents to gather and synthesize information from multiple sources in parallel.

Documentation:
See Agents: Multi-Agent Creation and Supported Architectures.

5. Ambient Agents with Triggers​

Overview:
Ambient agents can now be automatically triggered on a schedule or in response to specific events, enabling continuous monitoring, proactive responses, and automation.

Key Capabilities:

  • Schedule agents to run at fixed intervals (e.g., hourly, daily).
  • Define event-based triggers (e.g., new file upload, external webhook).
  • Automate routine monitoring, alerting, or data processing tasks.

Example:
Set up an ambient agent to monitor a data source and notify admins when new documents are added.

6. New Agent Architectures​

Overview:
Multiple advanced agent architectures have been added, supporting a wide range of AI use cases:

  • Plan and Execute: Agents autonomously plan multi-step tasks and execute them sequentially.
  • Reflexion: Agents iteratively reflect on their outputs, improving results through self-critique.
  • Research: Agents conduct in-depth research, aggregating and synthesizing information from diverse sources.
  • Codeact: Agents generate, execute, and refine code to solve problems or automate workflows.

Example:
Use the “Plan and Execute” architecture for a project management agent that breaks down goals and tracks progress.

Documentation:
See Agents: Supported Architectures.

7. Beak AI Gateway Integration​

Overview:
Unified access to over 100+ leading LLMs via Beak AI Gateway, with advanced routing, load balancing, and failover for high reliability and flexible model orchestration.

Key Capabilities:

  • Select from a wide range of LLM providers and models.
  • Configure routing rules and fallback strategies.
  • Automated load balancing for high-availability deployments.

Example:
Route general queries to OpenAI GPT-4, and fallback to Anthropic Claude in case of provider downtime.

Documentation:
See Beak AI Gateway.

8. Advanced Security Integration​

Overview:
Integrated Content Safety and Beak AI Security features provide robust protection and compliance, including automated filtering and detection for enhanced security of conversations and data.

Key Capabilities:

  • Real-time content filtering for sensitive or unsafe material.
  • Automated detection of compliance violations.
  • Integration with enterprise security policies.

Example:
Automatically redact PII from agent conversations before storing logs.

Documentation:
See Beak AI Security.

9. Observability and Monitoring​

Overview:
In-depth observability integration with advanced tracing, observation, and evaluation options. Monitor agent reasoning, actions, and outcomes for continuous improvement.

Key Capabilities:

  • Trace agent decision paths and action sequences.
  • Visualize agent performance and outcomes.
  • Evaluate agent responses against custom metrics.

Example:
Use observability dashboards to identify bottlenecks in a multi-agent workflow.

Documentation:
See Beak AI Observability.

10. Comprehensive File Type Support​

Overview:
The platform now supports all major file types as datasources, including documents, spreadsheets, presentations, images, and more.

Key Capabilities:

  • Ingest and analyze PDFs, DOCX, XLSX, PPTX, CSV, images, and more.
  • Extract structured and unstructured data for agent use.

Example:
Upload mixed file types for unified analysis.

11. Context Retrieval Enhancements​

Overview:
Improved context retrieval algorithms deliver more accurate data extraction from uploaded files, with inline citations and links to original files for enhanced traceability and compliance.

Key Capabilities:

  • More precise context matching for agent queries.
  • Inline citations in agent responses, linking to source files.
  • Enhanced compliance and auditability.

Example:
An agent’s answer about a policy document includes a clickable citation linking to the original PDF.

12. Multi-File Upload​

Overview:
Batch upload functionality enables users to add multiple files simultaneously, streamlining data onboarding and improving workflow efficiency.

Key Capabilities:

  • Drag-and-drop or select multiple files for upload.
  • Automatic batching and progress tracking.

Example:
Upload an entire project folder of documents in one action.

13. Chat Widget Deployment​

Overview:
Beak AI Studio can now be deployed as a browser-based chat widget, enabling seamless integration with external solutions and web platforms.

Key Capabilities:

  • Embed the chat widget in any website or web app.
  • Customize appearance and behavior to match branding.
  • Connect widget to specific agents or workflows.

Example:
Add the Beak AI chat widget to your company’s support portal for instant AI-powered assistance.

14. Model Controls in Playground​

Overview:
Users can now select different LLM models and adjust temperature parameters directly within the playground, enhancing experimentation and customization.

Key Capabilities:

  • Switch between available LLMs on the fly.
  • Adjust temperature to control creativity and determinism.
  • Compare model outputs side-by-side.

Example:
Test the same prompt with different models and temperature settings to optimize results.

15. Custom Skill Creation​

Overview:
Expanded options for creating custom skills, including REST API integrations, LLM prompt-based skills, and dynamic Python code generation/execution.

Key Capabilities:

  • Build REST API skills for external service integration.
  • Create prompt templates for LLM-based skills.
  • Write and execute custom Python code as agent skills.

Example:
Create a REST API skill that fetches real-time stock prices, or a Python skill that processes uploaded CSV data.

Documentation:
See Skills: Custom Skills.

16. Role-Based Access Control (RBAC)​

Overview:
Fine-grained RBAC enables administrators to assign specific access and privileges for agents, datasources, and skills, supporting enterprise-grade security and governance.

Key Capabilities:

  • Define user roles and permissions at a granular level.
  • Restrict access to sensitive agents, datasources, or skills.
  • Audit user actions for compliance.

Example:
Limit access to HR datasources to users with the “HR Admin” role.

17. Automated Deployment with Terraform​

Overview:
Streamlined deployment options now support automation using Terraform, accelerating setup, configuration, and scaling in diverse environments.

Key Capabilities:

  • Use provided Terraform modules for rapid infrastructure provisioning.
  • Automate updates and scaling operations.

Example:
Deploy Beak AI Studio and all dependencies to AWS using a single Terraform script.

18. Pre-Built AI Agents​

Overview:
A suite of ready-to-use AI agents is now available, designed to jumpstart operational use cases and demonstrate platform capabilities.

Included Agents:

  • CloudOps Agent: Automates cloud infrastructure monitoring and management.
  • ITOps Agent: Handles IT support tasks and incident response.
  • Service Desk Agent: Provides automated helpdesk support.
  • Internal Knowledge Agent: Answers questions using internal documentation.
  • And more…

Example:
Deploy the Service Desk Agent to handle common employee IT requests automatically.

📢 Feedback and Support​

Your feedback is invaluable! Please share your experiences and suggestions to help us continue improving Beak AI Studio. For support or to report issues, contact the Beak AI core team.