The future of work needs more than an AI chat

An invoice arrives by email. Someone opens the attachment, checks the details, looks for the matching order and passes the document on for approval. Later, the information is transferred into another system.

Every single step is ordinary. Together they produce interruptions, waiting time and sources of error.

An AI chat can help make sense of the invoice. But who moves the case forward afterwards? Who accounts for responsibilities? Who makes sure the right person decides?

This, as we see it, is where the next stage of AI lies: not only delivering information, but supporting connected work.

In the ISG Provider Lens® « Future of Work Services » study for Switzerland (2026), ISG describes how Swiss companies are moving beyond hybrid working models and bringing AI-supported collaboration, digital operations and workplace design closer together, and how firmly the question of AI governance has moved to the foreground.

A note on the source: the study is a provider comparison assessing 39 service providers. The version available to us comes from Swisscom, which is named a Leader in it. Hegyon is not part of the assessment. We refer to the market direction it describes, not to any evaluation.

At Hegyon we are concerned with what this development means for the daily work of Swiss SMEs.

From a helpful tool to a working process

Many companies have now gathered their first experience with AI. Texts are written faster, documents can be summarised and information prepared more easily.

That is valuable. But it does not yet solve the problem between the individual steps.

Operational work rarely consists of a single task. It requires information, responsibilities, decisions and a traceable path onwards.

A typical example:

• A customer enquiry arrives by email.
• Information about the customer sits somewhere else.
• Details needed to handle it are missing.
• A responsible person has to decide.
• The result has to be documented and passed on.

If AI improves only one of these steps, the handovers often stay exactly as they were.

The decisive question is therefore not « where else can we use AI? » but « which process should actually work better as a result? »

That also changes the measure of good software. What counts is not the number of AI features, but the relief in everyday work.

Human control belongs in the process

The more an AI takes on within a workflow, the more important clear boundaries become.

With a draft text, an unsuitable suggestion is usually corrected quickly. With an invoice, a personnel record or a financial decision, the requirements are different.

Here it has to be clear:

• Which information may the system use?
• Which steps may it prepare or carry out?
• When is human review required?
• Who may grant an approval?
• What happens when details are missing or data contradicts itself?

For us, human control is therefore not an addition bolted on at the end. It belongs in the design of the process.

The AI prepares. The human checks and approves where responsibility and risk demand it.

That does not mean every small thing has to be confirmed by hand. It means designing automation deliberately, to suit the case in question.

What this means for Swiss SMEs

Swiss SMEs do not need additional technology that first creates new work. They need solutions that take their actual requirements into account.

That includes local financial processes as much as comprehensible responsibilities, a traceable approach to data and an introduction that fits the business.

A platform can cover a great deal functionally and still miss everyday reality, if staff cannot see how they are supposed to work better with it.

We therefore consider three questions particularly important:

1. Does the solution fit the actual process? Not only in the ideal case, but also with exceptions, missing information and the queries that follow.
2. Is the handling of data traceable? A hosting location alone does not yet answer which data flows where, which providers are involved and who gains access.
3. Can the improvement be demonstrated? A quick demonstration is not yet evidence of better operations. What matters is what works under real conditions.

How we look at this future with Hegyon

Our direction with Hegyon is clear: we want to bring documents, information and operational processes closer together, rather than placing AI beside them as an additional island.

The Swiss focus is deliberate. Requirements such as QR invoices, VAT and local business procedures are part of daily life for our audience, and therefore part of how we think about the product.

Our guiding idea is: You approve. The AI does the work.

Behind it lies the notion that staff should spend less time searching for, transferring and passing on information. Their attention is needed where experience, judgement and responsibility are decisive.

Take the supplier invoice again. A sensibly designed AI-supported process should not end with reading the document. It should bring the relevant information together, make uncertainties visible and prepare the case for the responsible person to decide.

That is our ambition for the direction of product development. Which steps can be automated in a specific deployment depends on the available functions, the connected systems and the approvals agreed.

Maximum autonomy is not the goal. Reliable relief is.

The right way in is one concrete process

A broad platform vision does not mean a company has to change everything at once.

In our view, a sensible start begins with a clearly bounded process: frequent enough to make a difference, and small enough to judge both effect and risk.

Before starting, companies should record:

• How much processing time does the case take today?
• Where do waiting times and duplicate entries arise?
• Which errors or queries occur regularly?
• Which decisions must stay with people?

After the introduction, the same questions count again.

Because faster data entry achieves little if more has to be checked or corrected afterwards. An honest assessment includes the total processing time, the quality and the cost of running it.

Our conviction: AI has to prove itself in daily work

The future of work will not be decided by more capable models alone. What matters is how well companies translate those capabilities into their processes.

The direction ISG describes is relevant to us for that reason: integration, clear responsibility and operational maturity are not side issues. They decide whether an interesting technology becomes a lastingly useful working tool.

With Hegyon we pursue that direction at the level of concrete business processes for Swiss SMEs.

Not another tool that demands attention. A platform with the ambition to bring work together and relieve people where it counts.

Which recurring process costs your company unnecessary time today? Let us start exactly there.