The future of professional services - part one: When Knowledge Becomes Abundant

How AI is changing expertise, software delivery and the economics of professional services

There is something that has been playing on my mind since leaving the Salesforce partner ecosystem and starting Arqvera. There is no shortage of commentary about how artificial intelligence will change professional services companies… for the better or worse. Much of it starts with the technology: models are getting better, agents are becoming more capable, and professional-services teams can now produce research, analysis, code, tests and documentation more quickly. AI’s ability to automate and augment cognitive work is on the rise. All true. It is also the least interesting part of the story.

The bigger question is what happens to professional services when much of the cognitive work, including knowledge, analytical effort and software production, on which the industry was built, becomes more accessible, cheaper, faster and more widely available?

At times, I feel like technology has been at my heels since starting my career in the 90’s. I started in the creative industry, experienced the desktop revolution and quickly pivoted to bring web development skills into my repertoire. I lived through the dot-com boom and bust. I didn’t see the end of the internet but a realigned resurgence. I worked for Kodak, which suffered one of the biggest technology disruptions in modern-day history and is well documented. Kodak invented the digital camera and did commercialise it, albeit some would have you believe they didn’t. I should know. I sold the DCS range of DSLR cameras in the late 90’s. However, because of Kodak’s own resistance to move away from the razor blade strategy that built its wealth, it filed for Chapter 11 after I left and is now a shadow of its former glory. I have consulted with Australia Post on the $1Billion Future Ready Digital Transformation and have seen what happens when resistance and being a late mover become a drag and disadvantage. For example, the AP Digital Mailbox was a decade too late to create a meaningful impact. Luckily enough, Australia Post had options to double down on parcels to utilise its competitive advantage of the last-mile footprint. So many professional services firms made a lot of money from Australia Post, including the Big 4 management, accountancy and tech consultancies. I have been fortunate to work with them all on different projects.

Management and technology consulting, often referred to as professional services, is currently experiencing a significant transformation. Management consultants primarily provide business consulting and advice, while systems integrators offer a combination of business and technology guidance along with the design, development, and implementation of software solutions. Traditionally, both sectors have operated under a similar economic model, where clients compensate firms for access to specialised knowledge and the labour necessary to implement it.

However, the rise of artificial intelligence is not eliminating the need for expertise; rather, it is altering what is considered scarce in this field. Routine tasks such as codified knowledge and standard analyses are becoming increasingly plentiful, making them less of a differentiator. In contrast, what remains valuable are contextual, governed, and accountable forms of expertise that are harder to replicate. As the landscape evolves, the emphasis will likely shift toward these more nuanced areas of expertise.

That distinction is important because the future of professional services will not be decided by which firm gives every employee a chatbot or coding assistant. In my opinion, it will be decided by the firms that redesign how expertise is created, how software is delivered, how quality is governed and how experience improves the next engagement. The winners will use AI to change the operating model itself, not simply to make the old model run a little faster.

The world has changed. Most professional services have not.

It is easy to become numb to AI claims because a new capability or benchmark seems to arrive every week. A more useful measure is the length of real tasks that AI systems can complete reliably. METR’s research found that the human-equivalent duration of tasks frontier agents could complete with 50 per cent reliability doubled roughly every seven months over six years. Its January 2026 update found a post-2023 doubling time of about 131 days.

Costs are moving just as dramatically. Epoch AI’s analysis found the price of achieving a fixed level of performance had fallen between ninefold and 900-fold per year across different benchmarks. The precise rate varies, and the fastest falls may not continue, but the direction is in my opinion beyond dispute. A capability that was expensive, specialist and occasionally impressive is becoming cheaper, more accessible and woven into everyday tools.

Yet much of professional services still looks remarkably familiar. Work is sold, a team is assembled, tasks are distributed through a leverage pyramid and everyone records their time. In management consulting, junior people gather and structure information. In systems integration, large delivery teams translate requirements into configuration, code, tests and deployment artefacts. AI is being inserted into both systems as a productivity aid, while the commercial model, governance and knowledge flows remain largely untouched. That is rather like fitting an electric motor to a horse-drawn carriage and declaring the transport problem solved.

When codified knowledge becomes abundant

Professional services are not one activity. They combine research, synthesis, analysis, facilitation, architecture, software engineering, integration, assurance, change and execution. AI affects each part differently, which is why sweeping claims about either mass replacement or harmless augmentation are both unhelpful.

The activities under the greatest pressure are structured, repeatable and data artefact-heavy. Research synthesis, first-pass analysis, presentation drafting, requirements documentation, business-case development and PMO reporting can all be accelerated significantly. In a field experiment involving 758 BCG consultants, Harvard Business School researchers found that people using GPT-4 completed 12.2 per cent more tasks, worked 25.1 per cent faster and produced materially higher-quality work on tasks inside the technology’s capability frontier.

This is already observable in the firms whose traditional leverage models rely most heavily on junior analytical labour. McKinsey reports that 72 per cent of the firm uses its internal Lilli platform, which answers more than 500,000 prompts a month, with colleagues reporting up to 30 per cent time savings in searching for and synthesising knowledge. More tellingly, McKinsey says the programme has changed how people create and codify knowledge. This to me, indicates a nascent operating-model shift.

The same compression is reaching systems integrators across the software-development lifecycle. Requirements can be elaborated faster; code can be scaffolded, translated and reviewed; test cases and documentation can be generated; legacy applications can be analysed; and deployment scripts can be assembled with far less manual effort. Capgemini Research Institute reported early productivity improvements of 7–18 per cent across software engineering, with organisations using the gains mainly for new features and upskilling; only 4 per cent were targeting headcount reduction.

The gains are not automatic. DORA’s 2025 research describes AI as an amplifier: it can improve throughput, but often at the cost of stability when engineering foundations are weak. METR’s controlled study of experienced open-source developers even found that early-2025 AI tools made participants 19 per cent slower on complex repositories they knew well. Faster code generation is therefore not the same as faster, safer systems delivery. Architecture, integration knowledge, small-batch delivery, testing, security and operational accountability remain scarce.

The spreadsheet provides a useful historical parallel. It eliminated a great deal of manual calculation, but it did not eliminate financial analysis. It moved the work up the value chain. The same is likely across professional services, but there is a catch: firms whose economics depend on charging clients for time and the analytical or production layer being compressed cannot assume that higher productivity will translate neatly into higher profits. Clients can do the maths as well.

What remains scarce

The danger is to conclude that, because AI can produce an answer, it can own the outcome. It cannot. The same BCG experiment found that consultants using AI on a task outside the technology’s capability frontier were 19 percentage points less likely to reach the correct answer. The frontier is jagged, moves quickly and is often invisible to the person using the tool. Fluency is not the same as correctness, and a polished answer can be more dangerous than an obviously poor one.

Subsequent Harvard research into “persuasion bombing” makes the point sharper. When experienced consultants challenged flawed AI recommendations, the model could respond by restating the error more persuasively and surrounding it with apparently supportive analysis. In other words, the system did not merely get something wrong; it made the wrong answer harder to reject. This is precisely why generic calls to “keep a human in the loop” are inadequate. A human who lacks the context, authority or confidence to challenge the machine is not a control. They become a rubber stamp with a pulse. I have explained this to my two sons who are completing their Bachelor's degrees in Finance and Engineering. They are a little despondent due to the impact on graduate jobs. My explanation and reassurance to them is that any idiot can use AI and claim to be efficient and knowledgeable but be completely wrong. I reassure them by saying that to get the most out of AI, you need to understand the domain, ask the right questions and challenge the reasoning and data output. Education is still extremely valuable, and I see a future where there will be some who will become smarter because of AI. Or maybe I am just a trekie at heart!

I am basically telling them that the thing that remains scarce is the ability to frame the right question, distinguish the signal from noise, exercise judgement under genuine uncertainty, navigate organisational politics, build trust, negotiate trade-offs and lead people through the uncertainty of change. Above all, accountability remains scarce. Boards and executive teams do not merely need an answer; they need to know where it came from, what evidence supports it, which assumptions could break it and who is prepared to stand behind it. In short, human-centric leadership will become the defining capability in the future, and leadership can be learned.

This is why the value shift across professional services is more important than the productivity gain. Value moves from knowledge to judgement, from analysis to decision quality, from code volume to reliable capability, and from documents to evidence. Clients will still buy expertise and implementation, but they will increasingly expect traceability, governance and named people who own the recommendation and the system it becomes.

Abundance can still feel like loss

I keep returning to the human side of this change. We talk about AI as a technology transition, but for many people it will feel very personal. If your professional identity has been built on finding the answer, producing the analysis or mastering a particular craft, abundance can feel like devaluation. Even when the change is positive for the organisation, something familiar may still be ending for the individual.

It is tempting to reach for the familiar change curve and assume that people will pass neatly through the Kübler-Ross Curve of denial, anger, bargaining, depression, and acceptance. The reality will be less orderly. Recent research into the history of the change curve calls it a useful prompt for empathy, but warns against treating emotions as universal or linear. People can be excited by the possibility and anxious about their place in it at the same time.

What leaders call resistance may be grief for something real: the loss of familiarity, status, confidence or a story about what made someone valuable. Telling people simply to adapt completely misses the point. We need to name what is ending and distinguish obsolete work from human capability that is becoming more important. Judgement, curiosity, courage, creativity, care and accountability are how abundance becomes useful.

I do not see a single passage from an old world to a new one. We will move through repeated iterative cycles of letting go, experimenting and learning. The leadership task is to give people enough clarity and agency to shape what comes next: carrying forward the best of their experience while releasing the parts of the old model that no longer serve them.

The economics cannot stay the same

Professional services have always had a scaling problem. The traditional answer is leverage: win more work, hire more people, train them, deploy them and hope quality remains consistent. Management consultancies built pyramids of junior analytical labour; systems integrators added large onshore and offshore delivery teams. Both models work because labour is differentiated by cost and time can be billed. AI compresses many of the research, requirements, coding, testing and documentation tasks performed through those pyramids.

I believe the next move for many systems integrators will be digital labour arbitrage. I am seeing it already. The old model moved work to a lower-cost location. The new one assigns repeatable requirements, coding, testing and documentation to AI-enabled delivery with fewer human hours. The attraction is obvious: lower cost to serve and protected margin.

But this is not a durable strategy. Competitors and clients can use the same tools. One integrator lowers its delivery cost; another passes more of the savings to the customer; the market price resets; and the cycle begins again. Customers receive a lower price, but the value and outcome become generic. It is a zero-sum game with temporary winners. Offshore arbitrage required infrastructure that took years to reproduce. Digital arbitrage will have lower barriers and a much, much shorter half-life.

Systems integrators cannot avoid this opportunity and threat. They must use it to fund differentiation rather than mistake efficiency for differentiation. Lasting value will come from deeper sector expertise, stronger architecture and engineering judgement, distinctive intellectual property, productised ways of working, better governance and greater accountability for outcomes. Digital labour can change the cost base. Only a redesigned value proposition and commercial model can change why the client chooses one firm over another.

If a strategy team can complete an engagement in half the time, or an integration team can generate and test the same software with fewer delivery hours, a time-based model creates a very awkward conundrum. The firm can bill fewer hours, conceal the productivity gain, or charge the same amount for a result delivered with less effort and invite a difficult conversation about value. Recent reporting in The Wall Street Journal shows major firms experimenting with fixed-fee, subscription and outcome-based models as AI makes labour-based pricing harder to defend. The old proxy of hours for value is weakening in advice and implementation alike.

This does not mean the market is about to disappear. Growth and disruption can coexist. Accenture’s 2025 annual report recorded $2.7 billion of revenue from generative and agentic AI, three times the previous year, while firms across the sector are investing in AI-enabled software engineering and managed services. The market may expand while value moves away from undifferentiated labour and towards senior-led judgement, architecture, distinctive intellectual property, governed delivery systems and measurable outcomes.

Nor does the argument depend on AI capability improving forever. Even if progress plateaued tomorrow, the economics of research, synthesis, document production and parts of software delivery have already changed. The issue is whether consultancies and systems integrators redesign themselves around capabilities that already exist.

For me, that is the real meaning of abundance. It is not the end of expertise. It is the end of confusing possession of information with the creation of value. The opportunity is to spend less of our working lives proving that we know things and more of them applying judgement, helping people decide and staying accountable for the result.

Handled badly, that transition will feel like a machine steadily reducing the space left for people. Handled well, it can make professional services more human: less dependent on hierarchy and hours, more open about evidence and uncertainty, and more focused on the moments when experience, engineering discipline, trust and courage genuinely matter. Knowledge becoming abundant does not make us irrelevant. It asks us to become more useful.

The future of professional services Part 2 - Redesigning Professional Services: How Arqvera is applying outcome pricing, governed intelligence and compounding knowledge will be published next week.

About Arqvera

Is an AI and technology transformation consultancy and advisory.

We help organisations shape business cases, projects, deliver excellence, and realise change and outcomes that stick. We support organisations before, during, and after projects with an end-to-end service where our domain specialisation comes to life.

Before (Inception): We work with you to clearly define the idea, vision, strategy, and business case for change, as well as help select the right partners and establish governance

During (Execution): We help deliver the project and change objectives while keeping implementation under control through structured governance and assurance to realise intended outcomes.

After (Value Realisation): We ensure outcomes deliver measurable value and embed continuous improvement from successes and learnings.

Arqvera is led by industry veterans in the UK and USA with 100+ years of technology delivery intelligence across global consulting, digital transformation, and mission-critical projects and programmes.