Lilli has been at McKinsey for just a year, but she has already become indispensable to many of her colleagues. Lilli is available at any time of day or night, never in a bad mood, and never takes a vacation – and she's not an ordinary employee. Lilli is an AI platform that McKinsey has been testing since last summer. „Lilli integrates knowledge from a wide range of sources, from both structured and unstructured data, into a single platform“, explains Erik Roth, who is based on the US East Coast and serves as the Global Head of Design, Development, and Deployment of generative AI at McKinsey. He is known as „Lilli's Father“.

The platform is designed for employees to quickly search the firm's knowledge base or prepare specific routine tasks. About a year and a half ago, a four-person project team began developing Lilli, and now more than 120 people are involved with part of their work time. Over 75% of employees have already tested Lilli since the second half of 2023, with many using the tool multiple times a day, reports Roth: „There have already been over four million prompt requests, with about 150,000 new ones added each week.“

The prompts reveal what issues employees are addressing with Lilli. The „slide generator“ is particularly popular: it allows Lilli to prepare presentation slides in McKinsey's design. However, a consultant must still handle the final touches: „Everything that goes out to a client must be reviewed by a person first“, says Roth.

Artificial intelligence is not only transforming internal processes in consulting, but is also highly sought after as a consulting topic among clients across various industries. Generative AI can be especially effective in document analysis. This is the focus of a joint venture that PwC established in June with the German AI startup Aleph Alpha. The joint venture, called Creance.ai, aims to assist with regulatory requirements.

The first area of application is the Digital Operational Resilience Act (DORA), an EU regulation that will come into effect in 2025 to make the European financial market more resilient to cyber risks. Creance aims to simplify the required risk management for third parties. This involves reviewing past contracts to ensure they comply with DORA – which could involve thousands of documents in larger organisations.

„The flood of new standards, the complexity of organisations, and the seemingly endless amounts of data combined with a severe lack of available experts make this a nearly impossible challenge“, explained Aleph Alpha CEO Jonas Andrulis in the joint venture announcement. AI can sift through these documents and flag potentially critical points for further examination by staff. According to Björn Viebrock, a member of the PwC Germany management team, the offering is set to expand into additional applications.

The number of AI use cases is increasing both within consulting firms and in client projects. However, the impact of AI on the consulting business model itself is less clear. What does it mean for fee agreements with clients if AI takes over certain tasks? Does a client still need as many consultants? And what will consultants do with their time when routine tasks are eliminated?

Few are concerned about negative scenarios, as shown by the results of the „Management Consulting in Germany“ study by Lünendonk & Hossenfelder. Only 13% of respondents think that clients might use AI offerings as a replacement for consulting services. Just 2% worry that AI could negatively affect profitability, for example, if services are delivered with fewer employees and fewer billable hours. Eight percent foresee job cuts.

This relaxed attitude may be because, at least so far, AI has not been successful in all areas. The strategy consultancy BCG, which also collaborates with Aleph Alpha, conducted an experiment last year with over 750 of its consultants worldwide, and the results are likely to be favorable for BCG. The consultants were tasked with solving specific problems, some with and others without the support of OpenAI’s GPT-4 application. One group was asked to be creative and outline new ideas for products and market launches. This group achieved significantly better results with AI support, according to BCG.

Another group was tasked with identifying reasons for poor performance of a company based on data and conversation notes – a standard task for consultants. But this group performed worse with AI assistance. The consultants sometimes fell for incorrect conclusions from the AI, BCG explained. Additionally, the responses and solutions were less diverse. The authors concluded that generative AI is still not suitable for all areas – but this could change as AI continues to develop. There is also a next stage of development for GPT-4, known as GPT-4o, which was introduced this year.

According to the Lünendonk study, three-quarters of participating consulting firms agree or strongly agree that AI as a new consulting topic could boost their business. The industry has set double-digit growth targets for the coming years. This may explain why few expect job cuts in consulting, although this might be different for back-office areas.

In consulting, AI may be used for analyses and other topics, stated Walter Sinn, Managing Partner of Bain & Company in Germany, who presented the results. Due to the planned growth, this is unlikely to lead to job cuts, but rather to a shift in capacity. Michael Meyer, responsible for Accenture Strategy in Germany, Austria, and Switzerland, also views automation primarily as an option for the back office. Meanwhile, many consulting firms are searching for specialists, particularly those with AI expertise.

Larger consulting firms have a much easier time expanding their AI capabilities and funding the necessary investments compared to smaller consultants. The German consulting market is highly fragmented – only the top 9 on the Lünendonk list have a revenue in the triple-digit millions. The need for investment in AI could further accelerate consolidation, said Helmut Ahr, spokesperson for the board at Horváth, in the discussion. Conversely, smaller consulting firms could gain efficiency and effectiveness with AI support.

The time consultants will no longer spend on text summaries or slide creation should be well invested in training. Only those who focus on employee development can scale successfully with AI, believes Michael Meyer from Accenture. At McKinsey, Erik Roth initially observed an effect he calls „prompt anxiety": employees were unsure how to address Lilli effectively. Targeted training helped reduce these anxieties. "Employees use Lilli twice as often after an hour of training as before“, says Roth.

Roth believes that a collaboration between humans and technology will become standard in consulting, with task division likely varying by area. „There will be areas where you combine as few people as necessary with as much technology as possible, such as in the back office or for routine tasks. But where specialized knowledge and technological expertise are needed, such as in client consulting, the number of employees is likely to increase.“