The Technology Behind DoubleDesk

DoubleDesk combines professionally developed software with AI built around the way small businesses work.

It helps you handle calls, build useful processes and connect the systems you already rely on.

From the technology we chose to the testing behind it, our focus is practical capability, ease of use and control over what the platform can do for your business.

Designed by small business technology experts

DoubleDesk’s experienced design team has worked together for more than 30 years, launched successful products and designed software used at five Olympic Games. They are experts in applying technology to small business, having run small businesses for the past 30 years and implemented technology for over 5,000 small businesses.

Our design team created DoubleDesk’s architecture by hand, using established, mainstream technologies including .NET 10, SQL, Microsoft Azure and Cloudflare. Their practical experience shows in the depth of DoubleDesk’s features, its ease of use and the attention given to the everyday details that matter when running a business.

DoubleDesk was not built by vibe coding, and we do not use AI to mass-generate its code. All code is developed to our design specifications by qualified, experienced developers and tested by an experienced quality assurance team. We use AI for automated stress testing and code reviews as part of that development process.

Data hosting and AI platforms

All permanent storage of your DoubleDesk business data stays in your designated Microsoft Azure data centre in Australia, the United States or Europe, based on your location.

For AI tasks, we select the platform best suited to the work required. DoubleDesk uses a combination of Google Gemini, Claude and OpenAI (ChatGPT), chosen for the quality of their AI and their security and privacy protections.

Only transient AI requests the individual requests sent to complete a task are routed through an AI provider’s global service. Your data is never used to train AI models and is never retained by AI providers. The provider may process a request in another location according to available capacity and performance. This helps keep the service reliable and responses efficient, while permanent data storage remains in your designated Azure data centre. Each provider’s security and privacy arrangements apply to the processing of those requests.

Battle tested before release

Before release, DoubleDesk’s testing included more than 39,000 real calls and 20,000 MCP server tasks. This testing sits alongside the work of our experienced developers and quality assurance team, supported by AI for automated stress testing and code reviews.

We also commissioned a substantial, independent security and penetration assessment by an external, qualified CREST-accredited auditor. Taking approximately eight weeks to complete, it was designed to identify security weaknesses and test the platform’s defences.

DoubleDesk achieved a 100% pass in that assessment. The auditor’s report is available to customers on request.

Vibe code Workers and Playbooks in your own words

DoubleDesk itself was not built by vibe coding. We have brought that conversational way of creating things into the platform so you can shape how it works for your business. We have trained the AI behind our Worker and Playbook builders to understand how small businesses operate, how owners think and what they prioritise from responding to customers to keeping jobs moving and getting paid.

Describe what you want using the same industry terms and everyday language you use with your team. You do not need to write code or learn technical terminology. The builders help turn your explanation into a process you can review and refine, using the business systems you connect and the actions you allow.

With the relevant connections and actions available and enabled, you could ask for processes like these:
A Worker for your morning business summary

“Each morning, summarise new Shopify orders and overdue Xero invoices, then email me a short action list through Outlook.”

A Worker is an automated task that can bring information from connected systems together and run on a schedule you choose. You decide what the summary should cover and who receives it.

A Playbook for handling quote enquiries

“When someone calls for a quote, ask what they need, check my Outlook calendar for suitable visit times and save the enquiry in Zoho.”

A Playbook gives your receptionist a process to follow during a call. You decide which questions to ask, what information to record and what happens next—for example, having your team confirm a site visit with the customer.

Put your favourite AI assistant to work in your business

Already use Claude, ChatGPT or another AI assistant? DoubleDesk’s MCP server lets you connect a compatible assistant directly to your DoubleDesk account. Ask questions about your business and get help with everyday tasks through a familiar conversation, with less switching between screens and copying information.

Ask for a summary of today’s calls, help identifying enquiries that need attention, or a draft reply to a customer. A busy trades business could review incoming enquiries between jobs, while an office manager could turn recent calls and messages into a practical follow-up list. Your assistant works with the information in DoubleDesk, within the access you give it.

With the permissions you choose, your assistant can also update job tickets, organise business information or help build the instructions your receptionist follows. You can ask it to help create a Worker an automated task in DoubleDesk for something such as preparing a daily summary. Explain the outcome you want in everyday language, then review and refine the result.

You control access through security keys. Think of each key as a digital visitor pass: you choose which parts of DoubleDesk it opens and what the holder is allowed to do. Give each assistant or adviser its own pass, choose an expiry date and cancel access whenever you need to.

Start a Worker from your other business systems

A webhook is a start signal sent from one system to another. In DoubleDesk, it lets connected software remotely start a Self Running Worker—a Worker designed to carry out a saved task on its own. This means work can begin in response to an event in another system, without someone opening DoubleDesk to launch it.

For example, a customer asks for a callback on your website. The form saves their name, phone number and enquiry in a connected system such as Zoho. Your integration can then send a webhook to start an outbound-call Worker. The Worker retrieves pending callback requests, queues calls to those customers and follows the conversation process you have set up. It can record the results back in your connected system as part of its saved instructions.

You choose the Worker’s process, connections and calling instructions, and enable its remote access with a private key. Your website developer or integration provider sets up the trigger and handles requests that need a later attempt while the Worker is busy or waiting between runs. The webhook starts the work; customer details come from the connected system.

DoubleDesk pricing model

DoubleDesk’s pricing model is designed to make costs easy for small businesses to understand and manage. Usage is deducted from your available minutes: phone calls are billed by the minute, and AI work by the second.

For inbound and outbound calls, usage includes the actual call duration plus a small amount of AI time for call analysis, the transcript and any follow-up actions.

For Workers, you pay for the time the AI is actively working. Time spent waiting for a connected system to process a request does not count towards that AI usage. For example, a Worker could ask Salesforce or Zoho for prospects whose contact date falls within the next two weeks and whose quote exceeds $1,000. The CRM’s own processing time does not add to your AI usage.

DoubleDesk records your usage and the minutes deducted on a timesheet, so you can review where your allowance goes. AI can occasionally make mistakes, so every customer also receives 60 complimentary minutes each month, on top of their plan, to allow for work that needs to be done again.

How and why DoubleDesk began

DoubleDesk began with the small businesses our founders were helping through their technology consulting work. As AI started to become useful in everyday business, those customers wanted help answering calls and completing useful business tasks, taking real work off their teams.

The need extended beyond the phone to routine tasks using the business systems they already relied on. These businesses needed something they could manage and adapt themselves, without technical expertise, and maintain cost-effectively over time.

We set out to find a product we could recommend. We trialled many options, looking for the right combination of ease of use, flexibility and practical features. We could not find one that brought these capabilities together in the way we believed small businesses needed.

So we decided to build DoubleDesk: a platform designed to help small businesses put AI to work across calls and everyday tasks, with the simplicity and control to make it part of how they run their business.

FAQs

DoubleDesk uses established, mainstream technologies including .NET 10, SQL, Microsoft Azure and Cloudflare. Our design team created the architecture by hand, drawing on years of experience applying technology to small business.

DoubleDesk’s design team has worked together for more than 30 years. They have run small businesses for the past 30 years and implemented technology for over 5,000 small businesses. Their track record includes successful products and software used at five Olympic Games. Qualified, experienced developers write the code to the design team’s specifications, and an experienced quality assurance team tests it.

Yes. An external, qualified CREST-accredited auditor carried out a substantial security and penetration assessment that took approximately eight weeks to complete. The assessment was designed to identify security weaknesses and test the platform’s defences. DoubleDesk achieved a 100% pass in that assessment. The auditor’s report is available to customers on request.

They can use the business connections and actions you enable. Examples include preparing a summary from Shopify and Xero, sending it through Outlook, checking a calendar or recording an enquiry in Zoho. The relevant connections and actions must be available and configured.

It is a connection that lets an AI assistant work with your DoubleDesk information and tools. MCP stands for Model Context Protocol, but you don’t need to understand the technology to use it. Once connected, you describe what you need in ordinary language.

Yes. DoubleDesk supports connections from Claude, ChatGPT and other compatible AI assistants or agents. Your chosen service must support MCP connections, and your account or organisation must allow them.

You choose which areas of DoubleDesk it can access and which actions it can take. Give it the permissions needed for its job. Information it retrieves is shared with the AI service you connect, so choose access with that in mind.

Yes, by starting a Worker configured for outbound calling. For example, a website can save a requested callback in your connected CRM, then trigger a Worker to retrieve pending requests and queue calls. Customer details must be saved in the connected system; they are not passed into the Worker through the webhook.

No. It's totally secure, and the data always belongs to you. It's never used by any third party.

Ready to Work Smarter with DoubleDesk?

Join the business community using DoubleDesk to save time, simplify admin, and deliver better service every day.