Hilan and AI: When the Positioning Layer Turns Into an Earnings Engine
Hilan is pushing AI into almost every operating layer, but the 2025 filings still do not show a standalone earnings engine. This follow-up argues that AI is already embedded in projects, Data products, and value-added modules, while Summit and Boost add an interesting capability layer that is still not material to the P&L.
The main article already made the broader case that Hilan does not lack growth engines. This continuation asks a narrower question: has the AI layer, now spread across the presentation and multiple operating sections in the annual filing, already become an identifiable earnings engine, or is it still mostly a positioning layer, a sharper value proposition, and preparation for the next growth leg.
The 2025 answer is still conservative. AI is already inside Hilan's product and service stack, but it is not yet sitting in the filing as a separate profit stream. At Ness it is wrapped into projects, managed services, cloud, Data, and SAP work. In software-product distribution it sits inside an expanding product shelf around Qlik, Talend, Auto ML, Application Automation, and Summit. In payroll and organizational systems it appears mainly as an added-value layer and deeper usage inside an existing client base. What is still missing is proof: dedicated revenue, attach rates, separate profitability, or even a clean indication that AI is already moving the group's profit mix in a way that is hard to miss.
That matters because Hilan is building a story that can work in two very different stages. The first stage is defense and expansion inside the existing client base. The second stage is a real earnings engine, one that creates better pricing, clearer recurring revenue, or structurally better margins. As of year-end 2025, most of the evidence still sits in the first stage.
Three Takeaways Before The Numbers
First: the disclosure itself suggests that AI's first economic effect should arrive through efficiency rather than through a new revenue line. The business-solutions segment explicitly defines the replacement of human services by AI as a critical success factor. The presentation adds that Ness went through reskilling in response to the AI wave. That is the language of productivity improvement and service-model defense, not the language of a software product that has already opened a new earnings category.
Second: the part of Hilan where AI is pushed hardest still looks service-heavy. The business-solutions segment, where cloud, Data, cyber, SAP, and customer projects are concentrated, generated ILS 1.747 billion of revenue and ILS 133 million of operating profit in 2025. That is a very large segment, 57.9% of group revenue and 36.0% of consolidated operating profit, yet even after another year of AI rollout its operating margin was still only 7.6%.
Third: capital allocation around AI and automation looks careful and option-like, not like a move that already reshapes the income statement. Boost was acquired at 33.33% with a call option to reach 100% in five years. Summit was acquired at 55% with options to reach 100%. In both cases the consideration is explicitly not material. That is a classic way to build capabilities, not to declare that the profit engine has already arrived.
Where AI Already Sits In The Business
The right way to read Hilan in 2025 is not to ask whether it has an AI story. It clearly does, and it is broad. The real question is which layer of that story has already become productized, and which layer still functions mainly as a sales tool, a retention tool, or an implementation wrapper.
| Layer | What already exists in the disclosure | What it means economically | What is still missing |
|---|---|---|---|
| Business solutions | AI integrated into cloud, Data, SAP, cyber, support centers, automation, and solutions for finance and healthcare | Most likely sold as part of projects, managed services, and integration work | No separate AI revenue, backlog, or margin disclosure |
| Software-product distribution | GenAI solutions, broader Data Warehouse, Data Catalog, Data Replication, Qlik Sense, Auto ML, Application Automation, Summit integration | This already looks like a product shelf that can be packaged, distributed, and supported | No split showing how much of that shelf already produces recurring revenue or high-margin support |
| Payroll, HR, and organizational systems | The presentation talks about AI-based adoption and integration alongside broader solutions and regulatory support | AI likely enters here first as upsell and as added value to existing clients | No disclosure isolating penetration, pricing, or profit contribution |
| Summit | Transcription, subtitles, translation, meeting summaries, real-time transcription, and text-content management | This is a clearer AI product layer with a defined use case | The deal closed only after year-end and is still not material |
| Boost | Management and marketing of Monday solutions | The reasonable read is a workflow and implementation layer that can carry automation and AI inside enterprise processes | No disclosure that Boost already creates a standalone or material AI engine |
That chart matters because it captures the gap between the story and the economics. In payroll, HR, and organizational systems, where Hilan benefits from a deep installed base and complementary products, the operating margin already sits around 28.3%. In business solutions, where AI appears in almost every activity description, the margin improved but still stood at only 7.6%. In other words, AI is not turning Ness into a software business yet. At this stage it is mainly strengthening Hilan's ability to sell more projects, a broader service wrapper, and more managed services.
Ness's client mix also explains why monetization is likely to move more slowly than the presentation suggests. In 2025, 67.7% of business-solutions revenue came from government and public-sector bodies, versus only 32.3% from business clients. The presentation highlights strong activity in the government, public, and defense sectors. That is a strong fit for cloud, Data, and regulatory projects, but it also means AI often enters through tenders, service expansions, and implementation inside large systems rather than through a clean and fast product sale.
On the other side, software-product distribution looks like the place where Hilan already has its clearest AI shelf. The company describes GenAI offerings, the Qlik and Talend combination, Data Catalog, Data Replication, Auto ML, Application Automation, and the early integration of Summit into products and services. That already sounds like a shelf that can be packaged, distributed, implemented, and supported. But there is an important footnote here too: the company itself says the licensing market is moving from one-off license payments to subscription pricing, and that in the short term this shift can actually lower revenue and profit relative to traditional software sales. So even if commercialization is progressing, the accounting effect may lag the business progress.
Why This Is Still Not An Earnings Engine
The presentation opens with a long AI framing section. It talks about an efficiency shock, clearing backlogs, hyper-personalization by vertical, a shift from code writing to solution architecture, and adoption trends by sector. That explains very well why management thinks the theme matters. It still does not explain how much money it already makes for Hilan.
The annual filing says the same thing in a different format. It returns to AI again and again as a demand driver, a success factor, and a layer inside cloud, Data, support centers, SAP, fintech, digital processes, and financial-sector solutions. But nowhere does it isolate AI revenue, AI client wins, ARR, dedicated backlog, or separate profitability. That is not necessarily a disclosure failure. It is simply a sign that the layer does not yet sit inside Hilan as a distinct business line.
More than that, the filing does not describe a market where AI automatically creates a protected rent. Quite the opposite. In business solutions the company says that more than 40 major entities provide information-systems services in Ness's market, competition is intense, and the growing availability of AI tools means that in some cases clients themselves can develop certain capabilities internally. That is not moat language. It is the language of a market where vendors have to run fast just to protect their position.
That leads to the key conclusion from 2025: AI is already improving Hilan's value proposition, but the cash still seems to be arriving mostly through older commercial models. More projects, more managed services, more implementation, more modules, more support. That can still be a very good business. It is simply not the same thing as a new earnings engine that can already be isolated in the filings and pointed to directly.
What Summit And Boost Actually Add
This is where Hilan's real construction plan becomes clearer than the presentation headline. Summit is the closest asset to a truly productized AI layer. Ness acquired 55% of it in February 2026, and the company describes a solution for transcription, subtitles, translation, and meeting summaries, including real-time transcription and text-content management. The presentation adds that it is an enterprise AI solution in Israel, built around a unique hybrid model, with meaningful business growth and expanding adoption among large organizations.
Summit's strategic meaning is fairly clear. It adds a content-and-voice layer that is easier to explain to a client, package, implement, and connect to existing products. This is no longer just an AI consulting conversation and no longer just a cloud-service extension. It is a clearer product proposition. But the filing also stresses that the deal size is not material, and it closed only after the balance-sheet date. So Summit matters more as a strategic direction than as proof of 2025 earnings.
Boost is quite different. The filing does not present it as an AI company at all, but as a company focused on the management and marketing of Monday solutions. Ness holds 33.33% with a call option to reach 100% in five years, and again the amounts are not material. The reasonable inference from that description is that Boost adds a workflow, implementation, and operational layer. In other words, the place where an enterprise builds process flows, assigns work, and connects people, systems, and automation. That is not a pure AI layer, but it is a layer that can turn insight or automation into actual organizational action.
Taken together, Summit and Boost tell a more interesting story than the presentation alone. Hilan is not only buying "AI." It is building a value chain: data, analytics, documentation, and content on one side; workflow, implementation, and operations on the other; and around both of them a wider wrapper of cloud, integration, regulation, and managed services. That is a smart capability build. It just has not yet turned, as of 2025, into a profit engine that can be measured on its own.
When The Positioning Layer Becomes An Earnings Engine
For this story to move into the next stage, three things need to happen together. The first is better revenue visibility. Not necessarily a new reporting segment, but at least some indication that part of sales, expansions, or recurring revenue is now clearly tied to AI products, AI services, or defined modules.
The second is a visible change in earnings quality. If AI really replaces labor hours, shortens project cycles, or improves pricing, that should start showing up in Ness margins, software-product distribution, or upsell rates in payroll and organizational systems. As long as profit remains wrapped almost entirely in the old structure, the story will stay interesting but not fully proven.
The third is proof of cross-sell. Summit needs to show up not only as an interesting acquisition but as a product entering Hilan's existing client base. Boost needs to show that it helps connect workflow and automation to a broader enterprise value proposition. If both layers remain small side deals, they will strengthen positioning without yet creating an earnings engine.
That is why the right 2025 read is not that Hilan is late to AI, but also not that it has already cracked the model. This is a capability-building year, not an earnings-proof year. The market can like the direction, and reasonably so. What is still missing is the point at which the filing stops talking about AI as a horizontal language and starts showing it as a number.