Trust, Legal, Consulting Firms

A trust company in Sierre, a law firm in Sion, an engineering practice in Brig, a consulting firm in Martigny: such outfits are everywhere in Valais, they employ several thousand people between them, and they are, I believe, the most exposed link in the canton's economic fabric. No sector is as present in the Valais economy, and as little discussed in terms of its transformation by AI, as the intellectual professions. Exposed in the sense of vulnerability: their trades are changing fast, and global players are competing with them as never before. Exposed also in the sense of opportunity, because it is precisely these professions that stand to capture a substantial share of the value AI produces, provided they act now, and in the right way.

This is also the textbook case of the competitive shift described in chapter 2. Here, the essay's economic thesis finds its fullest expression.

One clarification before going further. My consulting work operates partly within this sector, and I work with Valais trust companies, law firms and SMEs on the very questions this chapter examines. That gives me an inside knowledge I would not otherwise have, but it obliges me to a certain discipline in writing: I will name none of my clients, promote no product, and draw from no analysis a recommendation that could be mistaken for a commercial pitch. The reader deserves to know where I am speaking from; that is now done. The arguments that follow stand, or fall, independently of this disclosure.

A dense, fragmented professional fabric

Valais counts several hundred trust companies, ranging from one-person outfits that keep the books for some thirty local SMEs to firms of fifteen to thirty staff covering the full spectrum: bookkeeping, tax, corporate advisory, wealth management, sometimes audit. It counts several dozen law firms, most of them also small. It counts a comparable number of engineering offices, architecture practices, and independent consulting firms. Add to this the notaries, insurance brokers, medical advisers, psychologists, coaches and trainers.

This fabric is fragmented and regional. Fragmented, because most of these firms employ fewer than ten people, many far fewer than five. That size fits the market they serve: a canton of three hundred and seventy thousand inhabitants, whose economic fabric consists mostly of SMEs that are themselves small. But that same size is also a source of vulnerability in the face of the changes now under way. A four-person trust company cannot dedicate a full-time position to technology watch, to evaluating AI tools, to the ongoing training of its staff. It cannot negotiate the pricing terms that larger firms obtain from software vendors. It cannot, on its own, build the technical infrastructure that would allow a rigorous deployment of AI tools while respecting professional secrecy.

Regional, because these professions do not serve a global market: they serve the businesses and individuals of the canton, sometimes of neighbouring cantons, more rarely beyond. Far from being a weakness, this regional dimension is precisely what gives these firms their value in their clients' eyes: they know the local fabric, master the canton's tax particularities, speak the territory's languages, including, in the Upper Valais, Walliserdeutsch, and are woven into relationships of trust built over years. This regional anchoring is exactly what global players cannot replicate, and it is probably the sturdiest lever on which these professions can build their future.

A transformation already under way

Generative AI is not going to transform these trades: it is already doing so, and the scale of that transformation is still underestimated by a good number of practitioners themselves.

Entry-level accounting work, data entry, coding, bank reconciliations, monthly statements, is becoming partly automatable, and that automation is accelerating as the tools improve. Standard legal drafting, simple contracts, template letters, memoranda on routine questions, can be considerably sped up by recent generative models, provided it is framed by rigorous professional review. Legal and documentary research, case law, doctrine, technical standards, tax references, drops from several hours to a few minutes for many everyday cases. Translating documents between the canton's languages, the federal languages and international languages is becoming almost free, at a quality level that, for many purposes, requires nothing more than a final read-through. Synthesising complex files, comparing proposals, preparing client presentations: all tasks that AI speeds up considerably.

This compresses billable time on intermediate tasks. If AI does in fifteen minutes what a staff member used to do in four hours, and if the quality of the result, after review, is equal or superior, then a significant share of the revenue built on those tasks erodes. Not immediately: client habits, billing arrangements, and trust in the tools do not change over a single quarter. But inexorably, over five to ten years. Firms that do not prepare for this will see their margins compress, and the smallest of them may not survive the squeeze.

More quietly, it is reshaping the skills required. The young staff entering these trades today should no longer be trained chiefly to execute the tasks AI now does better than they can; they should be trained to direct, to validate, to interpret, and to carry what the machine cannot: client relationships, situated judgment, professional responsibility, the defence of interests. This shift in skills cannot be decreed. It requires revising curricula, internships and evaluations, and that is not a project that gets done overnight.

The shift, seen from a Valais trust company

It is in this sector, more than in any other part of the canton's fabric, that the competitive shift takes on its full weight, and it is worth looking at what actually happens at the scale of a real firm.

Take an average Valais trust company: eight staff, a portfolio of sixty to eighty clients, revenue between one and two million francs. Over the past twenty years, this firm has watched its market erode. Standardised software solutions, the large online accounting platforms, mostly American or German, have absorbed a growing share of local SMEs' simple needs, which now handle their basic bookkeeping themselves for a few dozen francs a month. At the other end, the large international firms have kept the complex mandates for themselves: regulated audits, sophisticated tax structuring. The regional trust company has held its ground on what it does better than either side: knowledge of the local fabric, long-standing relationships with SME owners, the ability to work in French or Walliserdeutsch on questions sitting at the crossroads of federal law and cantonal particularities. But this niche has been shrinking, and margins along with it.

Generative AI, in this landscape, changes two things. It lowers the barrier to entry on complex mandates that were previously out of reach for a regional firm. And it raises the threshold above which regional, tailor-made service becomes competitive again against standardised solutions.

On the first point, access to complex mandates: the eight-person firm can now handle files that used to require fifteen or twenty staff. A comparative analysis of tax scenarios for transferring a family business, which once took a large firm several person-days, can now be prepared in a few hours by a senior partner who knows how to orchestrate the right tools. Structuring an intra-group loan between a Valais company and its French subsidiary, which used to require the backing of a large Zurich or Geneva firm for its international dimension, becomes accessible to the local trust company, which now has command of the relevant cross-border reference material. Drafting a legal opinion on a specific point of cantonal law, which once kept an assistant busy for two days, now comes out in a first draft within two hours and is revised in two more by the senior lawyer. The productivity multiples cited in chapter 2, four to one in development, five to one in design within my own group, are consistent with what trust companies and firms that have seriously engaged with the transformation are beginning to observe, even if the precise figure varies by mandate. The shared finding: with an equivalent human architecture, productivity per senior professional multiplies, without loss of quality.

On the second point, the return of tailor-made service: the same trust company can now serve needs that had drifted toward standardised solutions. The local SME that switched to an American management platform for a fraction of the cost of a traditional trust company discovers, as its needs grow more complex, that the platform cannot handle its specific reality. Its particular collective labour agreement, its eligibility for a cantonal support scheme, its family shareholding structure, its dealings with the bourgeoisie, a traditional Valais communal landholding body, from which it leases a plot: all local realities that no standardised tool can handle without the mediation of regional expertise. Where the trust company could not, yesterday, offer tailor-made service at the platform's price point, it can today, because its own productivity has been multiplied by the very tools that made the platforms successful in the first place. Tailor-made service becomes economically competitive again, at prices it had not been able to offer for twenty years.

This effect, added up across a few thousand comparable firms throughout the canton, changes the slope of the Valais professional fabric. It does not reverse urban concentration, the large firms of Zurich and Geneva will remain what they are, but it reopens, for regional firms, markets that had slipped away from them for a generation. That is the competitive shift, seen from the ground in Valais.

The architecture of the senior professional

This shift does not happen on its own. The productivity multiples mentioned above do not fall from the sky simply because someone installs an AI assistant on a staff member's desktop. They only materialise when a senior professional, someone who knows how to architect a problem, orchestrate production, and verify final quality, steers the use of the tools. This can be called the architecture of the senior professional, and it is what makes the difference between a trust company that modernises and one that multiplies its output.

This architecture carries precise strategic implications. Senior partners become more valuable than ever, which invites firms to concentrate on a small number of highly equipped senior professionals rather than grow by hiring juniors, and to retain expertise for as long as possible through part-time arrangements, mentoring and gradual handover. At the scale of the canton, the competitive shift requires a critical mass of senior professionals capable of playing this role, and their retention and attraction is the subject of chapter 13. It is in the intellectual professions that value is being reconfigured fastest, and so it is here that this demographic question presents itself with the greatest urgency.

Competition from global players

Alongside this internal transformation and this potential shift comes an external pressure that many Valais practitioners, I believe, underestimate, because it has not yet shown its face directly. The large international platforms, search engines, conversational assistants, integrated productivity tools, now offer individuals and businesses services directly that used to belong exclusively to the intellectual professions. An SME owner with a simple tax question can, in five minutes, get a plausible, free, immediate answer from a conversational assistant. That answer is not always correct, it may be incomplete or misleading, and a competent trust company remains indispensable for questions that carry professional liability. But a portion of demand, precisely the simplest and most profitable portion, is leaving firms without anyone noticing.

This pressure is all the more diffuse for not taking the form of visible competition. No major player is setting up shop in Sion to compete with local trust companies. But usage is shifting quietly, one query at a time, and the value it used to consume locally is migrating up to global platforms that, for the most part, pay no tax in Switzerland. This is not a catastrophe; it is a phenomenon measurable over a ten-year horizon.

Alongside this diffuse pressure, for the firms most exposed internationally, business law firms, consulting practices, comes direct competition from global players deploying sophisticated AI tools that regional firms could not develop on their own. In this segment, the balance of power does not favour Valais firms, and if nothing changes, it will not come to favour them.

The competitive shift therefore has a precise scope. It covers ground reclaimed on intermediate production from offshore players and generic SaaS platforms, the historic competitors of regional firms. It does not, however, touch the share captured by the hyperscalers (the major players of the global cloud) and the large platforms, which absorb, by sheer scale, the standardised, instant, global part of the market. The strategy of Valais firms must therefore be two-pronged: capture what comes back through the shift, on production specific to the territory, and defend what could otherwise drift away to the platforms, through the added value of situated judgment, trust, and sovereignty.

Professional sovereignty and professional secrecy

Yet in this difficult equation there is one asset that the regional intellectual professions possess and that global players cannot replicate: the combination of professional secrecy, territorial anchoring, and long-term trust.

Professional secrecy is not a corporate affectation. It is a condition of practice: a trust company knows its clients' financial situation, a lawyer knows the conflicts they are going through, a doctor knows their vulnerabilities. This knowledge is protected under Swiss law, and its handling is governed by strict rules. It is also protected, in practice, by a professional ethic that underpins client trust.

The arrival of AI raises an unprecedented question for professional secrecy: what happens to protected data when it is processed by tools whose infrastructure sits abroad? Does a trust company that uses an American AI assistant to draft a tax memo about its client send, without always realising it, information covered by professional secrecy onto servers subject to American law? The precise legal answer depends on the tools, the contracts, the terms of use, and it is more complex than is generally said⁴⁰. But the direction is clear: deploying AI in the intellectual professions cannot happen without serious thought given to the sovereignty of the data being processed.

This is where regional anchoring, paradoxically, becomes an additional commercial asset, one that compounds the competitive shift. A trust company that can guarantee its clients that the tools it uses comply with Swiss law, that their data does not leave the territory, and that the chain of accountability remains under Swiss jurisdiction, offers a service that global players cannot easily match. The argument does not hold for every client; some, indifferent to the question, will choose the cheapest option. But it holds for those who carry economic weight and have something to protect, which is to say, for the most profitable share of the market.

On this front, Valais holds a card that is not given to everyone. The canton has regional public operators, in energy and digital infrastructure, capable of hosting infrastructure placed under Swiss law and sized for the needs of the regional intellectual professions. It has training institutions, HES-SO Valais, Idiap, partnerships with EPFL and the University of Lausanne, that can contribute to developing suitable tools. It has, finally, a professional fabric dense enough for pooling resources to be economically viable. These three conditions are rare. They are not, as far as I know, being exploited to anything close to their potential.

A scenario for the decade

If nothing changes, here is what seems most likely to me over the next ten years. A fraction of firms, the largest, those best equipped with in-house technological skills, those best positioned on high-value segments, and those that manage to retain or recruit the senior professionals capable of orchestrating this work, will make the transition successfully. They deploy AI tools, automate what can be automated, move upmarket on strategic mandates, capture the competitive shift across their existing client base, and preserve or improve their profitability. Another fraction, the mid-sized firms that have neither the scale to invest, nor the specialisation to defend premium margins, nor the senior professionals to steer the process, will be gradually marginalised, will watch their revenue erode, and will eventually merge, be sold, or disappear. Today's fragmentation shrinks through consolidation, and the canton quietly loses part of its regional professional fabric. Nothing spectacular or dramatic: a diffuse movement one observes over ten years and comes to regret over fifteen.

If explicit decisions are made, a different scenario can unfold. The canton, in partnership with its professional associations, its public operators and its research institutions, sets up a shared infrastructure that gives firms of every size access to sovereign, reliable AI tools, compliant with Swiss law, at a reasonable cost. This infrastructure does not replace practitioners' expertise; it multiplies it. It allows a four-person trust company to access capabilities that would otherwise remain out of reach. It protects professional secrecy. By anchoring locally produced value here rather than letting it drift up to global platforms, it preserves a regional professional fabric that is, as we saw in chapter 3, one of the pillars of the resident skilled human capital this essay identifies as the central question of the decade.

Alongside this technical infrastructure comes a human dimension, perhaps even more important. The firms that succeed will be those that have retained and trained their experienced senior staff to orchestrate the new tools. This is not the same training as that given to juniors learning to use conversational assistants. It is training in structuring problems, orchestrating production, and critically validating outputs. As far as I know, this training is not systematised anywhere in Valais today. It could be, at modest cost, through a partnership between HES-SO, the professional associations, and a handful of advanced practitioners. This is a project the alpine campus mentioned in chapter 12 could carry, provided it is willing to treat it as a strategic priority.

This second scenario is not out of reach. It requires coordination, among Valais Economic Promotion, the professional associations, and the training institutions, that is not yet organised but could be within a year or two if the political will emerged. It requires investments that are modest against a cantonal budget, but structurally significant for the firms that would benefit from them. Above all, it requires a vision: the idea that the regional intellectual professions are not a sector like any other, but a strategic link in the canton's economic fabric, whose health determines that of the hundreds of local SMEs that depend on it.

A question of practical sovereignty

One thesis becomes clearer to me as I write this essay, and it strikes me as defensible even though it is unusual.

Digital sovereignty, if that word means anything at all at the scale of a canton, is not decided mainly at the level of large infrastructure. It is not decided in data centres; those choices are made at scales that far exceed the size of Valais, as we will see in the next chapter. It is decided, more modestly and more effectively, in the everyday practices of the regional intellectual professions. When a trust company in Sion uses tools compliant with Swiss law to handle its clients' files, that is practical sovereignty. When a law firm in Brig deploys document assistance that does not send its research up to foreign platforms, that is practical sovereignty. When a valley doctor keeps patient records on Swiss rather than American infrastructure, that is practical sovereignty. The sum of these individual and professional choices constitutes, at the scale of a canton, a more real sovereignty than all the talk about abstract digital sovereignty.

This practical sovereignty is precisely the one Valais can build, because it has the actors, the institutions, and the legitimacy to do so. It does not depend on federal or European decisions; it depends on choices that can be made here, starting now, by the professions concerned, with the support of cantonal public authorities. This is, I believe, the one project the canton can act on without needing anyone's permission or an outside partnership.

And it is, incidentally, the condition for the regional professional fabric to make it through the decade without eroding. This is a rare alignment: a decision that serves the general interest, the canton's identity, and the future of several thousand skilled jobs, all at once. Alignments of this kind do not come along often. When one does, it does not wait.