M1 Plugin Fl Studio

[Michele Firmasyah]

Theweird thing: the JUCE tutorial say to only check VST3. When I then build it spits out the error message. I have now created a new project with standalone checked, when I compile the plugin actually opens now.

Really sorry for the late reply, had a lot to do and then forgot about it, its a shame.

BUT: it works, both a demo project and my own. I have now updated to the very last version of JUCE.

Now when I compile, the plugin is immediately in ableton (before I weirdly had to manually search for it which sometimes made everything just crash) and loading it works perfectly fine.

I've created two Copilots 1 and 2. Only 2 is tied to a web page for data source. I've created a custom topic and a conversational plugin for both copilots. In the topic/plugin, I've added a Generative answers node after the trigger phrase node. In the generative answers node, I do not add any data source but instead add my context prompt inside the properties section of the node under the custom instructions section. Also, in the node, I mark the input as Activity.Text

I've tested this topic/plugin in both copilots but after providing the input to copilot, it does not reply back with anything and when I look at the flow of the topic/plugin, it is marked as completed.

Additional context, in my chat with copilot - I provide it details of a random project and in the context prompt, that I already provided, I ask the copilot to extract the facts from my message and reply back in a professional style.

thanks @HenryJammes - that's the confusing part. I'm using our company website as the data source. Configured under the main "Generative AI" menu, which I can see is the same as the configuration for the "Create Generative Answers" action. I've given the action a complex prompt which includes background and context for the task and constraints around the output, e.g. "Keep your answer brief and limit the number of references to three (3) or less." . Sometimes this works for a given input, other times it doesn't and drops through to fallback topic.

@biscuits1234, @DougM_Barhead this is by design. Generative answers only work when a data source is configured, because the generated answer needs to grounded in the input content the data source will return.

I have a very similar issue. I've setup a custom prompt for "Create Generative Answers" which sometimes works when responding to a number of set test phrases, and other times the topic doesn't trigger at all and just drops through to the Fallback topic

Hi @NoamBendelac,

When I tried my vst3 plugin in FL for first time I have also some problems. Unfortunately I do not remember exactly a scenario, but now everything works. I think I had two problems:

Maybe try to check definitions of JucePlugin_Manufacturer and JucePlugin_IsSynth in AppConfig.h. If JucePlugin_Manufacturer correctly set, then try to add breakpoint in juce_VST3_Wrapper.cpp for constructor of JucePluginFactory (line 2669):

A plugin is a reusable piece of code that can perform a specific task or provide specific functionality for a copilot. For example, a plugin can help a copilot answer a natural language query, execute a workflow, connect to an external system, or provide topic-specific guidance. A copilot is a conversational or UX-based assistant that helps users accomplish their tasks and goals in a specific domain or application.

A plugin, in this context, is created in the plugin authoring experience in the Microsoft Copilot Studio. This feature allows users to create and edit plugins using a graphical user interface and publish them to the plugins registry.

The plugins registry helps you create a plugin once and use it in multiple copilots. The registry provides storage and management for metadata and execution information for plugins. Users can apply the power and flexibility of plugins to enhance the capabilities of copilots without writing code for each copilot separately. The various copilots interact with the plugins registry to discover plugins and execution information available for a user. This capability enables AI plugins to be created once and reused many times.

Prompt plugins are plugins that can answer natural language queries from users using natural language processing (NLP) and natural language generation (NLG) techniques. For example, a prompt plugin can answer a question like "What are the sales opportunities for ACME Corp?" by querying data in Microsoft Dataverse and generating a natural language response that can be customized (such as a bulleted list).

Flow plugins are plugins that can execute multi-step workflows using Microsoft Power Automate. For example, a flow plugin can create a new contact in Microsoft Dynamics 365 after manager approval or send an email to a customer with a prescribed template.

Connector plugins are plugins that can connect to external systems or data sources using Microsoft Power Platform connectors. For example, a connector plugin can connect to Salesforce or SAP using Power Platform connectors and retrieve or update data. Connector plugins can be certified connectors that are shared across tenants or environment-specific custom connectors.

Topic plugins are single-turn conversational threads between a user and a copilot that can be created to answer a specific user utterance. For example: a topic about store hours with a trigger phrase check store hours can return the store hours. These topics can still handle user utterances such as "see store opening hours" due to the power of LLMs.

The plugins registry stores and manages the plugin metadata and execution information. The registry is a single source for discovery of plugins authored from Microsoft Copilot Studio. The registry helps you discover integrated apps available at the tenant level used in the Microsoft Admin center by Microsoft 365 Teams. The registry shows you plugins available to a user, or used by copilots in Microsoft Copilot Studio. A plugin has the same security as its underlying artifact, such as a flow and can provide a customized list of plugins for a specific user role.

This sample set of the various copilots integrates with the plugins registry to consume plugins. These copilots include Microsoft copilots for Dynamics 365 apps, Microsoft copilots for Teams, and custom copilots authored using Microsoft Copilot Studio. The list is expected to grow in the future as more first-party and third-party copilots are developed.

The administration layer is where administrators can use the Microsoft Admin Center to choose the integrated apps available for their tenant. Administrators can allow, deploy, or block apps for specific or all users. An integrated app can have multiple plugins. The apps available from Microsoft Copilot Studio-created plugins are retrieved from the plugins registry and include:

Dynamics 365 and Microsoft Copilot Studio: Includes Dynamics 365 plugins from teams such as Microsoft Dynamics 365 Sales and Microsoft Dynamics 365 Field Services. You can create custom prompts and topic plugins.

Connector specific: A sample of Salesforce and SAP plugins are called out here. Every certified connector plugin created by Microsoft or ISVs shows up as their own integrated app for fine-grained management.

The plugins available to a maker include plugins assigned through deployed integrated apps. The plugins available to a maker are discovered through the plugins registry. Makers are free to further customize the plugins based on how they want to use the plugin flyout experience. The plugin execution information gets retrieved from the plugins registry and includes the runtime path.

Plugins that are Dataverse-based: Microsoft Dynamics 365 plugins that access or modify Dataverse data, Microsoft Copilot Studio topic plugins, and prompt plugins. This path supports Microsoft Entra ID authentication. The calls are made to the underlying system on behalf of the user.

Plugins that use the Power Platform connector ecosystem: This path helps execute connectors that provide access to internal and external systems of record. It supports explicit configuration of credentials per user such as Salesforce connection credentials that are then maintained per user.

This layer represents the various data stores and systems where the data and business logic reside. These stores and systems include Sales, Field Service and other Dynamics 365 products, Microsoft Copilot Studio, Power Automate, external systems such as Salesforce, SAP, and others.

As a prerequisite for all flows, an administrator uses the Microsoft Admin Center to configure the corresponding integrated apps and assigns them to users, including the Copilot user referenced here. In these flows, we first see the configuration phase for a copilot user in Copilot for Teams. After the user signs in, the plugins applicable for that user are retrieved from the plugin registry. The plugins in a user's flyout belong to the integrated apps their administrator configures for them and the ones they have access to in the plugin registry. The user can then configure the plugins they want to use in the copilot experience in the flyout in their Copilot for Teams experience.

In the runtime flow for the Dynamics 365 plugins, the orchestrator maps a copilot user's utterance of Get opportunities for ACME corp to a set of candidate plugins. A plugin from Dynamics 365 is found to be best candidate and then executed, translating the user's utterance to SQL against the dynamic data the user has access to. The results are then returned to the user. Teams copilot can additionally add data from Microsoft 365 and summarize the results.

In the runtime flow for Copilot Studio prompt plugins, the user's utterance is again mapped to a plugin. In this case, the matching plugin is a prompt authored in Microsoft Copilot Studio. The prompt grounds data in Dataverse table orders and calls into OpenAI to summarize the results for order 123 and returns the summary to the user.

3a8082e126