DC Agentiqs Ver.1.2 Release Notes

DC Agentiqs Ver.1.2 expands support across a wide range of areas, including AI reasoning, long-running agents, working folders, document processing, and external tool integration.

This page summarizes the main updates by purpose. For detailed setup and usage, follow the links to the relevant feature pages.

info

The latest models from each LLM provider are supported. Apply them to your provider settings using Update models in bulk.

Overview of additions

Skills Reuse business rules and expertise across agents and workspaces.	Mission mode Lets an agent continue working autonomously until the goal is reached, with progress checks and resume after interruption.	AI reasoning support Adjust the reasoning level in five steps for complex analysis, design, and coding tasks.
Entra ID integration Connect to Azure OpenAI / Microsoft Foundry with a company account and simplify initial setup for organizations.	Work folders & system tools Specify multiple folders and use file operations and command execution from chat.	Improved import of existing design documents Converts graph-paper Excel sheets, AutoShapes, sequence diagrams, and ruled PDF tables into a form that AI can reference more easily.
Local LLM support Use Ollama or LM Studio and keep your data from being sent outside the organization.	Dynamic sub-agents Delegate investigation and work to multiple specialized agents and run them in parallel or in the background.	Next Design AI Toolkit Supports Next Design introduction, migration, and extension work with AI.

Checks before upgrading

If you want to keep using settings created in V1.1, review the following changes first.

File Access tool settings

Change	V1.1 and earlier	V1.2 and later	Required action
Path specification used by tools	Relative and absolute paths	Absolute paths only	Convert any relative paths to absolute paths
Function name	Keyword Search	File Content Search	Update prompts that still use the old name

Workflow save behavior

Change	V1.1 and earlier	V1.2 and later	Required action
Save / discard method	Confirmed with [Publish], discarded with [Reset]	Saved/restored with Version History	For workflows in shared workspaces, the displayed content is carried over as-is. Back up if needed before upgrading.

Default behavior changes

• Skills are now enabled by default. Existing agents are also treated as enabled. If you do not need them, turn them off in the agent's "Skills" tab.
• Reasoning is now enabled by default for reasoning-capable models. This improves quality, but may increase token usage and response time. You can adjust the reasoning level from the chat model selector.

Compatibility with older versions

• The save format for shared workspaces has been updated. If you manage shared workspaces with Git, we recommend that one representative user open each page once in the new version, commit and push the changes, and then have other users pull them before starting work.

caution

Please standardize the Agentiqs version used within your team.

Details of new features

AI models, reasoning, and cost management

Feature	What changed / benefit	Details
AI reasoning	You can now adjust reasoning levels for GPT-5, Claude 4, Gemini 2.5 and later models. This helps achieve more accurate responses for complex analysis and design tasks.	Set the reasoning level
Local LLMs	You can use Ollama or LM Studio as AI providers. This lets you process sensitive data without sending it outside the organization.	Use a local LLM
Context usage and manual compression	You can check usage directly on the chat screen and compress conversation history whenever needed.	Check context window usage
Provider priority and model purpose	You can change provider order and assign models to different purposes such as summarization, sub-agents, and image generation.	Set provider priority and model usage
Model information and cost management	You can update model information in bulk while checking differences. During chat execution, you can monitor cumulative cost and token usage.	Update models in bulk / Check usage cost
Prompt cache	Prompt-structure optimization makes it easier to benefit from caching even in long conversations.	Check AI model usage cost

Agents and chat experience

Feature	What changed / benefit	Details
Skills	Business rules and expertise are automatically detected and used. Dedicated tabs, app-wide folders, and references to Skills from Claude Code, GitHub Copilot, and Codex are also supported.	Use Skills
Web search and code interpreter	You can search for the latest information and perform calculations or processing based on script execution results.	Basic AI agent settings
Mission mode	The agent can plan, execute, report progress, and finish tasks autonomously. Interrupted tasks can be resumed from the same chat.	Advanced AI agent settings
Dynamic sub-agents	Multiple investigations and tasks can run in parallel or in the background, and the results are collected by the parent agent.	Advanced AI agent settings
Create and continue conversations from chat	The first prompt used in "Create with AI" is carried over to the AI chat in the edit screen, so you can continue the conversation from creation to refinement.	Create with AI
Quick reply	You can answer AI questions using choices or number keys. Question sets that combine multiple questions are also supported.	Use quick reply
Tool-call display	You can more easily see the operation, target, state, and result in chat.	Check tool-call details
Chat organization	You can group chats and pin frequently used chats.	Group chats
HTML live preview	You can view and interact with AI-generated HTML in an isolated environment.	Live preview HTML code
Conversation breakpoints	You can exclude older conversations from being sent to the AI and lighten the context while keeping settings intact.	Set conversation breakpoints
Icons and colors	You can make agents, tools, and knowledge easier to distinguish by role and purpose.	Set the agent appearance

Workspaces and working folders

Feature	What changed / benefit	Details
Multiple working folders and system tools	You can specify multiple folders per chat and use file operations and command execution. Permission modes let you control the scope of operations.	Set working folders
Sharing working folder settings	Folder names, descriptions, and permissions are retained per path and can be reused in other chats.	Set working folders
Sensitive file guard	Sensitive files such as .env and private keys are blocked from reading by default. No additional settings are required.	Can I handle sensitive files without security risks?
Workspace ordering	You can reorder workspaces by drag and drop.	Reorder workspaces
List organization	You can display agents, knowledge, tools, and workflows in flat, grouped, or folder views, and pin them.	Organize lists

Workflows

Feature	What changed / benefit	Details
Version history	You can save versions with names (including AI-suggested names and descriptions) and restore earlier states when needed. Up to 20 versions are kept per workflow.	Save and restore workflow versions
Scheduled execution	You can automatically run workflows on an hourly, daily, or weekly schedule.	Run automatically on a schedule
Workflow debugger	You can isolate problems with breakpoints, step execution, and variable inspection/editing.	Debug workflows
Input variables	Now supports nine data types, including strings, numbers, booleans, arrays, selections, date/time, and files/folders.	Define input and internal variables
Questions	You can build fixed-response checks into a workflow using single-select or multi-select questions.	Configure question steps
Execution history and cost	You can check elapsed time, title search, variables after completion, and total/step-by-step costs.	View execution history
Continue from create to edit with AI	The first prompt used in "Create with AI" is carried over to the AI chat on the edit screen, allowing you to continue from creation to adjustment.	Create with AI

Knowledge and document processing

Feature	What changed / benefit	Details
Office documents and PDF import	Graph-paper Excel sheets, ruled PDF tables, and Office AutoShapes can now be imported in a structured form.	Import Office documents and PDFs
Images inside documents	You can interpret images inside Office documents and PDFs by using multimodal-capable models.	Interpret images in documents
PlantUML conversion for sequence diagrams	Sequence diagrams drawn in graph-paper Excel or AutoShapes can be converted to PlantUML.	Import sequence diagrams
Index update notifications	You can check which knowledge items have re-parse enabled and how many files are targeted, then update only the necessary files.	Check index update notifications

Tools and external integrations

Feature	What changed / benefit	Details
App-wide MCP server settings	A registered MCP server can be used across all workspaces. Import of existing settings is also supported.	Configure MCP servers globally
Image generation from chat	You can generate concept diagrams and image ideas directly in chat and download them.	Generate images in chat
Lightning Review	When adding a finding, you can set type, severity, priority, status, process, date, and comments together. Rejection history (reason for finding, correction policy, etc.) can also be retrieved.	Lightning Review tool
Next Design	You can automatically explore related models and locate target models by model ID (GUID). Automatic diagram layout is also supported.	Next Design tool
File Access	Supports multiple edits, file-content search, and file-name search using Glob patterns.	File Access tool

Security and admin features

Feature	What changed / benefit	Details
Entra ID authentication	You can connect to Azure OpenAI / Microsoft Foundry with a company account, without distributing API keys individually.	Connect with Entra ID authentication
Automatic Entra ID client ID setup	Embedding the client ID in enterprise licenses simplifies initial setup for end users.	Set up an Azure app for Entra ID authentication
Organizational shared settings	AI provider settings and other items can be encrypted and distributed, and changes after the first registration can be reflected automatically.	Distribute via organizational shared settings

Other UX improvements

Target	Improvement
Create screens	When creating agents and workflows, it is now easier to choose between gallery, manual creation, and Create with AI.
Agent settings	Changes to context and tools are now reflected directly, and tool settings can be collapsed for review.
Icon picker	The Apply button is shown only when an icon name is entered.
Knowledge settings	When knowledge is added to an agent, [Use knowledge] is enabled automatically.
Approval dialog	The full command is wrapped for display, and choices can be selected with buttons. Waiting for approval is preserved even if you switch screens.
Question / approval popups	The display size is more compact, and multiple questions are now shown one per page.
Chat display	The title of a new chat and the running cost are reflected before response completion.
List screens	NEW is shown for newly created agents, knowledge, tools, and workflows.
In-app tutorials	Tutorials have been added so that first-time users can confirm the workflow.
Help	Related help pages can now be opened from each screen.

Fixes

• Improved target selection for Lightning Review integration Fixed an issue where the wrong target could be selected if multiple review files or outlines with the same name existed in the same workspace.
• Improved child-entity retrieval for Next Design integration Fixed an issue where agent calls to retrieve child entities could fail.

In addition, various other stability and usability improvements have been made.