Acudesk brings together conversational AI, system integrations, and powerful automation tools — and we know you may have a few questions before diving in. Below you’ll find answers to the most common topics our customers ask about, from agents and datasets to pricing, setup, and API integrations. If you don’t find what you’re looking for, our team is ready to help.
Actions
Yes. Each action has its own access control. Even if an agent has access to an action, it still respects user-level permissions.
You provide examples, keywords, and intent descriptions. The agent uses this information to choose the correct action based on the user’s request or API input.
Actions define what an agent can do (e.g., create a ticket, update an order, send an email). They map to your API endpoints and allow agents to perform real operations.
API
Yes. You can specify “allowedSchemas” to narrow what the tool should detect.
Yes. A JavaScript SDK is available for embedding conversational agents or integrating API features into your systems.
Yes. Everything—agents, document processing, transcription, schemas—can be accessed through our REST API.
Audio and voice transcription
When possible, the model differentiates speakers, making transcripts easier to read and analyse.
Yes. It works well with background noise, outdoor recordings, mobile phone audio, and “quick voice notes.”
No. Audio files are deleted immediately after the API call.
Billing
No. Acudesk is fully month-to-month or top-up as needed. You can upgrade, downgrade or quit anytime.
You purchase a monthly credit pack. Credits are deducted as you use tools, agents, and APIs. Packs include bonus credit depending on tier.
Standard uses a smaller, cheaper model. Plus uses a more capable model for complex tasks. Auto selects the best model per request for balanced performance and cost.
Conversational agents
Yes—if you give them permission. Using datasets and actions, agents can read information, update records, or trigger workflows securely.
Absolutely. You can define tone, confidence, formality, response style, voice, and supported languages. Each agent can feel unique to your organisation.
Unlike generic chatbots, Acudesk agents are connected to your datasets, grounded in your terminology, and able to execute real actions. They provide accurate, context-aware answers instead of guessing.
Datasets
Yes. You can connect MySQL, MariaDB, PostgreSQL, or MSSQL using read-only credentials or custom views.
They can—but only if you allow it. Each dataset can be granted or restricted per agent.
You can connect SQL datasets, API datasets, document collections (RAG), and grounding datasets. Each provides different ways for agents to answer questions or retrieve information.
Document processing tool
No. Once your API call completes, all input is immediately discarded.
Yes. It can recognise pages belonging to different schemas—like invoices, contracts, and delivery notes—all inside a single uploaded file.
No. It uses visual AI interpretation, which works with scans, photos, handwritten notes, stamps, and mixed layouts.
General
Yes. Most customers use Acudesk for internal automation, but agents can also be embedded into client portals, Teams, or web apps for customer-facing support.
Not necessarily. Most configuration is point-and-click, and you only need SQL if you want to train database datasets. More advanced features like actions and integration agents benefit from basic API knowledge.
Acudesk is an AI platform that connects your business data, tools, and workflows to intelligent agents and automation engines. You can link datasets, define actions, configure grounding, and deploy conversational or integration agents that work directly with your internal systems.
Grounding
Yes. With API grounding, Acudesk requests fresh context directly from your systems before generating a response.
Yes. Both types use grounding to understand your terminology and interpret inputs correctly.
Grounding connects the agent’s understanding to your real business data—clients, locations, product names, IDs, departments, internal terms—so responses are context-aware and accurate.
Integration agents
Yes. They can chain multiple API calls, apply rules, run conditional logic, and orchestrate complex workflows all within one request.
You configure a custom JSON schema. The agent always produces structured output that matches the format your system expects.
Integration agents process machine-to-machine requests. They read input, apply your business rules, call actions, access datasets, and return structured JSON—without any chat interaction.
Privacy & Security
No.
AI agents operate under customer-defined rules with human oversight and safeguards in place.
No.
Acudesk does not sell personal data or PHI.
Yes.
Acudesk supports GDPR requirements, including data access, correction, deletion, and lawful data processing with appropriate safeguards for international transfers.
More information on GDPR compliance can be found under the Data Processing Agreement (DPA) section of our website.
Yes, for healthcare customers.
Acudesk acts as a Business Associate and signs Business Associate Agreements (BAAs). PHI is handled according to HIPAA requirements. More information on HIPAA compliance can be found in the Data Processing Agreement (DPA) section of our website.
No.
Acudesk does not use customer data to train AI models by default. Training occurs only with explicit customer permission. Protected Health Information (PHI) is never used for training without legal and contractual authorization.
Only the data sources you explicitly connect and authorize, such as internal documents, databases, or business systems.
You do.
Data can be deleted at anytime when no longer needed.