Summary

• Ares Intelligence has raised approximately £2.5 million from Creandum, Lakestar, and angel investors associated with Google DeepMind.
• Its software conducts confidential employee interviews in more than 70 languages to generate prioritised automation recommendations.
• At industrial customer Tremonia, Ares says its assessment identified more than 400 hours of potentially avoidable manual work per month.

London-based Ares Intelligence has raised approximately £2.5 million to expand software that interviews employees and identifies opportunities for AI and automation within existing business processes.

The investment was backed by Creandum and Lakestar, with participation from angel investors associated with Google DeepMind. Ares intends to use the capital to develop its product, recruit additional employees, and expand its commercial operations across Europe and the United States.

Founded in 2026 by Matt Falconer and Alfie Ranstead, the company offers an AI interviewer that conducts confidential voice conversations with employees in more than 70 languages. It analyses their descriptions of everyday work to identify repetitive tasks, inefficient handovers, and activities that could potentially be improved through automation.

The findings are combined into a prioritised implementation plan for management teams. Rather than beginning with a particular technology product, the platform collects evidence about operational problems before recommending areas for investment.

Employees describe how work actually happens

Organisations planning AI investments often begin by examining processes documented by senior managers or technology departments. Those records describe the intended workflow but may not capture the adaptations employees have made to complete tasks using systems that do not exchange information effectively.

A department may regularly copy data between applications, for example, because an integration was never built or an earlier process has become outdated. Employees can recognise the additional effort involved even when the activity is absent from formal process documentation.

Other inefficiencies arise when information is transferred between teams. A finance employee may need to request details from operations, while customer-service staff repeat work already completed elsewhere because the necessary records are not readily available.

Ares uses its interviewer to collect these experiences across an organisation rather than relying only on workshops with a small group of managers. The resulting conversations can reveal patterns that would be difficult to identify from one department’s perspective.

Its platform then examines the responses and produces recommendations ranked according to estimated opportunity. These can include applications of AI, conventional software automation, or other changes intended to reduce unnecessary work.

Distinguishing between those approaches is important because some problems can be resolved by connecting existing systems rather than introducing a language model. A business may obtain greater value from improving data quality or changing an approval process than from automating a task built around unreliable information.

Industrial customer identifies manual workload

Ares has described an engagement with Tremonia, a business separated from Mercedes-Benz, in which its software interviewed employees across different functions, including senior management and people working on the factory floor.

Within four weeks, the company says the assessment identified more than 400 hours of potentially avoidable manual work per month. The finding provides an indication of the opportunities identified during the project, although it should not be confused with hours already eliminated through completed automation.

Manufacturing organisations often rely on several systems for production planning, inventory, engineering, procurement, and financial reporting. When the systems are poorly connected, employees may use spreadsheets, emails, and informal handovers to keep work moving.

These workarounds can be effective in the short term because they allow employees to handle exceptions that existing software was not designed to manage. Over time, however, they can create repeated effort, limit visibility, and introduce errors as information moves between departments.

Interviewing people across the organisation can help establish where the same underlying problem affects several teams. A manual report produced in one department may exist because another system does not provide information in a form suitable for its users.

Once such relationships are identified, management must determine whether addressing the problem is commercially worthwhile. That involves considering the time spent on the activity, the cost of implementing a change, and the operational consequences of altering the process.

Confidentiality affects the quality of the evidence

Employees may be reluctant to discuss inefficient practices if they believe their comments could be used to assess individual performance or identify mistakes. The quality of an organisation-wide assessment therefore depends partly on how participation and confidentiality are managed.

Ares says its platform offers controls for removing names, voice information, and identifying role metadata before findings are presented to management. It also provides deployment options involving customer-controlled infrastructure or its own services in European and American environments.

Those arrangements allow businesses to consider how the technology fits their information security requirements, although organisations remain responsible for assessing privacy obligations and the suitability of the selected deployment.

Confidentiality can be more complicated in small teams, where a distinctive role or description of a process may identify an employee even when a name has been removed. Businesses must therefore consider which information is collected, how results are aggregated, and who can access individual responses.

Clear communication about the purpose of the interviews also matters. Employees may be more willing to describe practical obstacles when they understand how the information will be used and what controls apply to the resulting assessment.

At the same time, interviews produce subjective evidence. Workers may estimate the time spent on tasks differently, or identify a problem without knowing the technical and contractual constraints that prevent it from being resolved immediately.

Recommendations based on those accounts need further examination before they become committed technology projects.

Prioritisation comes before implementation

Ares positions its platform as a way to reduce the work required to discover where AI could provide a return. The service combines conversations with analysis and recommendations, allowing management to consider a range of opportunities without organising a lengthy manual research exercise.

The resulting priorities can help organisations decide which workflows deserve technical investigation. They do not establish that an automated solution has already been designed, tested, or integrated with the systems involved.

After identifying an opportunity, a company may need to examine existing software, data permissions, integration requirements, and the controls required to maintain reliable operation. Some improvements will be relatively straightforward, while others involve changes to several applications or departments.

Costs also need to be assessed against the expected benefit. A process that consumes several hours of manual effort may not justify expensive custom software, whereas a smaller improvement repeated across thousands of transactions could provide a more substantial return.

The platform’s value will therefore depend on the accuracy of its estimates and the extent to which customers can turn recommendations into effective changes.

Funding supports a larger commercial test

Falconer and Ranstead previously operated automation business Instruct, where customers encountered difficulties identifying which processes should be automated first. Ares was developed to address that initial discovery stage.

The company says it has secured enterprise customers in Europe and the United States during its first months of operation. The latest financing will support additional product development and commercial expansion.

Published figures for the investment vary when expressed in euros. Contemporaneous reporting identifies amounts around €2.6 million and €2.9 million, while several sterling accounts place the round at approximately £2.48 million to £2.5 million. The article therefore retains the approximate sterling amount rather than presenting an exact euro conversion.

No valuation or breakdown of individual investors’ commitments has been disclosed. The financing provides resources to expand the business, while the longer-term commercial test will involve whether its assessments generate reliable opportunities and repeat customer demand.

As organisations commit more money to AI, identifying where the technology fits existing work may become as important as selecting individual products. Ares is seeking a role in that preliminary assessment, with its future growth depending on the quality of the evidence collected and the commercial results customers achieve after implementation.