Emily: Building a Virtual Assistant That Actually Gets Things Done

Over the last three years, artificial intelligence has gone from a niche technology discussed primarily by researchers and technology companies to one of the most talked-about business topics in the world.

Almost every week brings a new announcement. A faster model. A more capable chatbot. A new benchmark. A new promise that AI will transform the way we work.

For many organisations, however, the reality has been somewhat less dramatic.

People have experimented with AI tools. They have generated emails, summarised documents, drafted reports, created presentations, and asked increasingly sophisticated questions. The technology is undeniably impressive and, in many cases, genuinely useful.

Yet despite all of this progress, most professionals still finish their day facing many of the same challenges they faced before AI arrived.

• Their inbox remains full.

• Meetings still need organising.

• Customers still require responses.

• Tasks still need tracking.

• Follow-ups still get forgotten.

The problem is not that people lack information. If anything, most organisations are drowning in information. The challenge is that information alone rarely creates outcomes. Knowing what needs to happen and actually making it happen are two very different things.

That observation became the starting point for Emily.

At White Rabbit Foundry, we spend a significant amount of time helping organisations improve the way they operate. Sometimes that involves data and analytics. Sometimes it involves automation, digital products, reporting platforms, AI solutions, or operational redesign. Every client is different, but a surprisingly consistent pattern appears across almost every engagement.

The greatest inefficiencies are rarely found in the core work itself. They are found in the activities surrounding it. Professionals spend hours coordinating meetings, responding to routine enquiries, chasing updates, managing calendars, preparing documents, organising information, maintaining task lists, and handling countless administrative activities that individually seem small but collectively consume enormous amounts of time.

These activities are rarely strategic. They are simply necessary. The challenge is that they compete for attention with the work that actually creates value. A consultant wants to focus on solving client problems but spends hours managing communication. A business owner wants to focus on growth but finds themselves trapped in administration. A project manager wants to focus on delivery but spends much of their day coordinating people and information.

The more organisations grow, the more these administrative layers tend to expand. Ironically, many productivity tools designed to reduce this burden often add to it. Each new platform solves a specific problem, but creates another system to manage, another notification stream to monitor, and another workflow to maintain. The result is a workplace that is increasingly connected but not necessarily more productive.

Much of the first wave of AI adoption focused on generation.

• Generate an email.

• Generate a report.

• Generate a presentation.

• Generate a piece of code.

These capabilities created enormous value because they accelerated activities that previously required significant manual effort. However, they also revealed an important limitation. Generation is only one step in a much larger process. An AI model can draft a meeting invitation, but someone still needs to organise the meeting. It can generate a proposal, but someone still needs to review, send, and follow up on it. It can summarise a conversation, but someone still needs to act on the decisions that emerge from it.

This is where the conversation around agentic AI has begun to gain momentum. Agentic AI refers to systems that move beyond simply generating information and begin helping users achieve objectives. Rather than answering a question and waiting for the next instruction, these systems can execute workflows, coordinate actions, monitor progress, and help move work forward.

In simple terms, traditional AI answers questions. Agentic AI helps complete tasks. While the technology is still evolving rapidly, we believe this shift will ultimately create more value than content generation alone. Businesses do not succeed because they have more information than their competitors. They succeed because they can act on information more effectively.

Emily began as an internal experiment. We wanted to understand what would happen if we focused less on conversations and more on outcomes. Instead of building another chatbot, we asked a different question.

What would a genuinely useful digital assistant look like?

• Not a demonstration.

• Not a technology showcase.

• Not a novelty.

• A real assistant.

The kind of assistant that could help professionals manage the small but important tasks that accumulate throughout a working day. As we explored this question, one thing became increasingly clear. Most people do not need another platform. They do not want another dashboard. They do not want another application demanding attention. They simply want help getting things done. This insight shaped every decision that followed.

Rather than forcing users to learn a new interface, we decided to meet them where they already worked. For most professionals, that place is email. Despite decades of attempts to replace it, email remains at the centre of modern business communication. It connects teams, customers, suppliers, partners, and stakeholders across virtually every industry.

Building Emily around email allowed us to remove friction and focus on usefulness. If someone can send an email, they can use Emily. The technology becomes invisible. The outcome becomes the focus.

Creating a virtual assistant sounds straightforward until you start trying to build one. Generating intelligent responses is no longer the difficult part. The challenge is everything that comes afterwards. People do not work in neat, predictable workflows. Every organisation has exceptions, nuances, preferences, and edge cases that rarely appear in process diagrams.

A meeting might need to be rearranged because someone is travelling. A follow-up might depend on a specific customer response. A reminder might need to change based on shifting priorities. Real work is messy. To support these realities, Emily had to become more than a language model.

Behind the scenes, the platform combines AI, workflow automation, scheduling logic, business rules, contextual memory, cloud infrastructure, and integration capabilities designed to support practical outcomes rather than impressive demonstrations.

Throughout development, we repeatedly tested Emily against real-world scenarios rather than idealised examples.

• Could it help manage a busy consultant’s schedule?

• Could it support a small business owner juggling multiple responsibilities?

• Could it reduce the administrative burden that often consumes valuable working hours?

• Could it provide enough reliability for users to trust it with important activities?

Those questions became far more important than benchmark scores or technical specifications.

One of the most exciting aspects of Emily is not any single capability but the way multiple capabilities combine to support everyday work. Consider a consultant managing several client engagements simultaneously.

Their inbox receives meeting requests, project updates, stakeholder communications, and new opportunities throughout the day. Each email may require a different response, a reminder, a follow-up action, or a scheduling change. Emily can help coordinate these activities, reducing the amount of manual administration required while ensuring important actions are not overlooked.

For small business owners, the value can be even greater. Many entrepreneurs operate as sales manager, customer service representative, administrator, marketer, and operations lead all at once. Their challenge is rarely a lack of effort. It is a lack of time. Emily can help manage routine enquiries, organise reminders, assist with communication, and support day-to-day operational activities, allowing owners to focus more of their attention on growth and customer relationships.

Even individual professionals can benefit from having a digital assistant that helps organise priorities, track commitments, coordinate meetings, and maintain momentum across multiple projects. These may seem like small improvements in isolation. However, anyone who has spent years managing a busy professional workload understands how quickly these activities accumulate. Saving a few minutes dozens of times each week becomes meaningful very quickly.

Perhaps the most surprising lesson from development was how differently users evaluated Emily compared to traditional AI tools. When people use a chatbot, they often focus on the quality of the response.

• Was it accurate?

• Was it helpful?

• Was it well written?

When people use Emily, they evaluate something entirely different. They evaluate trust.

• Will it remember?

• Will it follow through?

• Will it complete the task?

• Will it behave consistently?

In many ways, internal users treated Emily less like software and more like a colleague. This observation fundamentally changed how we thought about the platform. Reliability became more important than creativity. Consistency became more important than sophistication. Completing a task successfully mattered more than producing an impressive answer.

The lesson was simple but important. People are not looking for AI that sounds intelligent. They are looking for AI that is useful.

After months of development, testing, refinement, and internal use, we are now opening Emily to a wider group of beta users. Over the next three months, participants will have the opportunity to explore the platform, experiment with different use cases, and help shape the next phase of development.

This beta programme is not simply about identifying bugs. It is about understanding how virtual assistants fit into real-world workflows.

• Where do they create the most value?

• Which tasks should remain human-led?

• Where does automation genuinely improve outcomes?

• What builds trust?

The answers to these questions cannot be found in a laboratory. They emerge through everyday use, real feedback, and practical experience. That is exactly what this beta programme is designed to achieve.

The most important story about Emily is not Emily itself. It is what Emily represents. For years, technology has focused on helping people access information. More recently, AI has focused on helping people generate information. The next stage may be helping people act on information.

That shift has the potential to change how individuals and organisations think about productivity, administration, and work itself. The future of AI is unlikely to be defined solely by bigger models, faster responses, or more impressive demonstrations.

It will be defined by usefulness.

• By whether these systems help people reclaim time.

• By whether they reduce friction.

• By whether they allow individuals to spend more energy on creativity, judgement, relationships, and strategic thinking.

That is the future we believe is worth building. Emily is our first step towards it. If you would like to be among the first to explore the platform and help shape its future, we invite you to join the beta programme at: emily.whiterabbitfoundry.com