The Operational Backbone: Why Mid-Market Needs a Company OS in 2026 (Not Another SAP)
A German mid-market machine builder kicked off a Systems Applications Products (SAP) S/4HANA Public Cloud migration in 2024. Eighteen months later, it was over. €780,000 (EUR) was gone, and the business kept running the way it always had: in Excel, in Germany's standard accounting platform (DATEV), in Outlook. The story is publicly documented by Knowledge & Performance Solutions (KPS).
What we see every week at IJONIS
Every week at IJONIS in Hamburg, the picture in first-call advisory looks the same. A tired Chief Executive Officer (CEO), a resigned head of Information Technology (IT), controlling still calculating in Excel workbooks with 30 tabs, broken cell references, and Visual Basic for Applications (VBA) logic only one person in the building still understands. The reflex answer is always the same: a bigger, newer ERP will sort it out.
Our take is the opposite. What mid-market really needs in 2026 is not an ERP migration. It is an operational backbone above the existing systems. We call it the CIAH (pronounced C-I-A-H).
"We are not migrating your SAP. The layer on top is what we are building. The one you really wanted, in 12 weeks instead of 18 months."
— Jamin Mahmood-Wiebe, Co-Founder, IJONIS Unternehmergesellschaft (UG) haftungsbeschränkt, Hamburg
The 2026 picture: spreadsheet economy meets S/4HANA shock
The data tells the story. According to the investment survey 2024 from the German-Speaking SAP User Group (DSAG), 88 percent of surveyed companies feel overwhelmed by S/4HANA conversion.
A parallel finding from Mittelstand-Heute 2025: 60 percent of active migrations come in over budget, over time, or both. SAP support for the older ERP Central Component (ECC) ends in 2027. The pressure is real.
So is the reaction: consultancies pitch million-euro programs running 18 to 36 months. At the end stand systems that don't make the business better, only more expensive.
Mid-market operations live in a reality that ERPs barely reach. Roughly 40 percent of corporate planning in the German Mittelstand still runs in Excel: workbooks with 30 tabs, grown over seven years, with VBA macros, broken cell references, and one person holding the whole thing together in their head.
The Lünendonk study 2025 names IT modernization and automation as the top two investment themes in mid-market. In parallel to a reality where operational processes still happen by hand.
The problem isn't that the ERP is too small. The problem is that operational work happens between systems, and no single system covers that layer. An inquiry comes in by Outlook, the quote gets calculated in Excel, the order lands in SAP B1, the confirmation goes out via WhatsApp Business, the books live in DATEV, and sales keeps a parallel list in HubSpot. Six systems, none of them talking to any other. Right there, in the gaps between systems, mid-market burns its margin.
The missing layer: above, not instead
What mid-market is missing isn't better accounting software or a bigger ERP. It is a standalone layer above the systems already in place, reading from SAP, DATEV, Outlook, Excel, and HubSpot but replacing none of them. Its job is to own the operational work that today happens in the gaps between systems. The layer has three core jobs:
-
Bring data together. From every system where operational work happens, draw a unified picture of the business. The customer with the open inquiry in Outlook is the same customer with a delivery delay in SAP, who also called via WhatsApp five days ago.
-
Spot what is off. The layer knows what normal looks like and flags what isn't. Fourteen hours of overrun on a project quoted at 80 hours. A complaint rate climbing for six straight weeks. A warehouse running thin on an A-class part.
-
Trigger action. Not just display, but act. The escalation email is drafted. The complaint case is opened in the ticketing tool. The reorder draft is queued at the supplier, one click away, with the last purchase price already filled in.
This is not a new ERP. Not a BI dashboard either. Nor a Zapier flow with a few connections. The CIAH is its own architectural layer, living between the systems, owning the operational work.
The classic build-vs-buy question for custom software shifts. The choice is no longer "which off-the-shelf product do we buy". The new question is the inverse. Which custom layer do we build, so that the off-the-shelf products we already own finally feel like one system?
What a CIAH is (and why we coined the term)
The CIAH layer is the answer to that gap. The term is coined on purpose: no existing category captures the combination of data plus intelligence plus action in a custom layer above the stack. "Company Operating System (OS)" and "Mittelstand-Betriebssystem" became hollow marketing labels in 2026. Microsoft positions Microsoft 365 as the "operating system for the Mittelstand", Sensified calls its platform "ai-os", Plotdesk and half a dozen other vendors claim the same label. None of them ship the combination we mean: data plus intelligence plus action in a custom layer above the existing stack.
Every letter is anti-positioning against an existing category.
The breakdown below shows what each letter argues against.
The Hub doesn't replace SAP, DATEV, HubSpot, or any spreadsheet. It sits above them. It pulls what it needs, writes back what it decides, and lets the operational systems handle the things they already do well. An S/4HANA project asks: "How do we replace what works?". A CIAH project asks the opposite: "How do we finally use what we already have?".
The four components of a CIAH
A CIAH consists of four connected components in practice: a data layer, a role-based daily home, an intelligence layer, and an action layer. Exact shape varies per project, the underlying pattern doesn't. Sequence and depth decide whether the Hub ends up showing data only or actually owning operational work day to day.
Data layer: connectors into every source system
Connectors into every relevant source system: ERP, Customer Relationship Management (CRM), scheduling, accounting, spreadsheets, mailboxes, messengers. From this emerges a unified data model of the business.
The unflashy part, but it decides everything. If the data isn't right, the best intelligence layer won't help. Anyone with experience in AI integration for ERP and CRM systems knows the rule: 60 percent of build effort sits in the first two components.
Role-based daily home for every team
Every role in the company, from leadership to sales to warehouse, gets its own view. Not the universal overview that everyone uses differently and nobody owns. A surface that fits the role. Leadership sees liquidity, order book, projects sliding off plan. Sales sees open quotes, pending follow-ups, at-risk customers. Warehouse sees critical stock, late deliveries, complaints.
Intelligence layer: KPIs, anomalies, forecasts
Key Performance Indicator (KPI) calculation, anomaly detection, basic forecasts. Classical statistics where they are enough. AI models where classical methods fall short. The point is: this layer isn't just display, it is the trigger for the next component.
Action layer: rules and agentic workflows
Here is what BI never does. Deterministic rules where the logic is clear. A simple rule reads like this in a CIAH:
WHEN inventory(item) < threshold(item)
AND last_order(item) > 14_days
THEN generate_reorder_draft(item, supplier, last_price)
Agentic workflows kick in where the logic is context-dependent: an inbound complaint should be reviewed, classified, routed to the right team member, and accompanied by a draft response.
This layer is where most competing products stop. They show data, they alert a human, then the human switches tools to act. In a CIAH, the action happens in the same surface.
Four real-world workflows
Four workflows from real engagements: manufacturing, field services, boutique agencies, direct-to-consumer e-commerce. Each follows the same pattern: aggregate data, detect anomaly, trigger action. The figures below are illustrative scenarios from typical mid-market engagements, not benchmark data.
Example A: DACH machine builder
A typical mid-market manufacturer in the Germany-Austria-Switzerland (DACH) region, 80 staff, custom-build orders, special machinery. Stack: DATEV for accounting, SAP Business One for inventory and order processing, Outlook for inquiries, Excel for quoting and project planning, WhatsApp Business for direct customer contact. Pain: inquiries get lost. A customer mails in, also pings on WhatsApp, sales is tracking two threads, eventually one drops off the radar.
| App | What happens | Where in the CIAH |
|---|---|---|
| Aggregate data | Outlook inbox, SAP B1 order and delivery records, DATEV open items, Excel quoting sheet, WhatsApp history per customer | Data layer |
| Detect anomaly | Open inquiry in Outlook plus old delivery delay in ERP plus last WhatsApp contact 5 days ago plus no sales reply for 48h | Intelligence layer |
| Trigger action | Escalation to leadership. Draft reply with current delivery status, referencing prior emails. Customer record updated. | Action layer |
| Outcome | Inquiries stop falling through. Sales has one inbox instead of five. | Role surface: leadership and sales |
Example B: Field services (commercial maintenance, R480k engagement)
A commercial maintenance provider with field technicians. Stack: Zuper for job scheduling, TaskR for techs in the field, QuickBase for the customer record, Sage for invoicing. Pain: jobs get signed off on site via a Certificate of Completion (COC), and then nothing happens for weeks. Invoicing falls behind, Quality Assurance (QA) stays open, cash flow suffers.
| App | What happens | Where in the CIAH |
|---|---|---|
| Aggregate data | Zuper job status, TaskR field reports including signed COC, QuickBase customer record, Sage invoice status | Data layer |
| Detect anomaly | COC signed but no invoice draft in Sage for more than 3 days. Or: QA stays open more than 5 days after closeout. | Intelligence layer |
| Trigger action | Agent drafts the invoice in Sage. Pings QA in Slack. Escalates to operations after 48h with no response. | Action layer |
| Outcome | Closes the gap between job completion and invoicing. R480k engagement, 12 weeks. | Role surface: operations and finance |
Example C: Boutique agency (services firm)
A small agency, 12 staff, project-based work. Stack: HubSpot for CRM and deal scope, Harvest for time tracking, Xero for accounting, Slack for internal comms. Pain: projects run over budget, nobody catches it in time, and 200 hours per quarter never get billed because no change order ever got written.
| App | What happens | Where in the CIAH |
|---|---|---|
| Aggregate data | HubSpot deal scope, Harvest actuals per project, Xero invoiced hours, relevant Slack channels | Data layer |
| Detect anomaly | Project at 120% of quoted hours (threshold set per project during Discovery Sprint, as of 2026), no change order in HubSpot, time entries still landing in Harvest | Intelligence layer |
| Trigger action | Agent drafts a change-order email with cost summary and hour log. The project manager (PM) gets the draft for sign-off in Slack. | Action layer |
| Outcome | Roughly 200 unbillable hours per quarter recovered. At €150 per hour: about €30,000. | Role surface: project lead and leadership |
Example D: E-commerce (Shopify-native brand)
A direct-to-consumer (D2C) brand on Shopify, seven- to eight-figure annual revenue. Stack: Shopify, Meta Ads, Google Ads, Loop or Returnly for returns, Klaviyo for email. Pain: individual products, called Stock Keeping Unit (SKU) in retail, have ruinous return rates, but marketing keeps scaling them because the Customer Acquisition Cost (CAC) looks fine in isolation. The reaction lands with the Chief Marketing Officer (CMO).
| App | What happens | Where in the CIAH |
|---|---|---|
| Aggregate data | Shopify orders, Meta and Google Ads campaign data, Loop return rates per SKU, Klaviyo cohorts | Data layer |
| Detect anomaly | SKU acquired via campaign X has 32% return rate (category threshold 18%, set during Discovery Sprint, as of 2026). Spend doubled in the last 14 days. | Intelligence layer |
| Trigger action | Auto-pause the Meta campaign. Draft a revised product page (clearer sizing, more honest imagery). Notify CMO. | Action layer |
| Outcome | Stop paying to acquire returners. Ad budget moves to profitable SKUs. | Role surface: CMO and ecom team |
Why this is possible in 2026: the build-cost collapse
What was a half-million-euro program with two years of runway in 2024 is now a 30,000 to 80,000 euro project shipped in 8 to 12 weeks. Three shifts drove the change.
First, AI-assisted app building. Tools like Retool AI ship working prototypes in 1 to 2 days. Retool's AI app builder lets non-developers build enterprise apps. Current mid-market AI builds run 30,000 to 150,000 dollars (USD) over 6 to 12 weeks, four to six times less than two years ago.
Second, Forward Deployed Engineers (FDE) as a delivery model. FDE postings are up 800 percent across 2025 and 2026. Accenture and Microsoft just launched a joint Forward Deployed Engineering practice, and Ernst & Young (EY) for the United Kingdom (UK) and Ireland did the same in April 2026.
The model Palantir made famous (small teams embedded with the client, shipping in weeks where classical consulting plans in quarters) just hit the mainstream.
Third, the "services as software" thesis. Sequoia argues openly that the next generation of software isn't selling tools, it is shipping outcomes. That is exactly what a CIAH does. The customer doesn't buy a toolkit. They buy a finished operational layer that runs from day one.
What a CIAH is not (anti-positioning)
The term is new, so the negative comparison is worth its space. A CIAH is none of the products below. Six adjacent categories (Agent OS, AI-native ERP, iPaaS, BI, Internal Tools, Robotic Process Automation (RPA)) each deliver part of the result. None ship the combination of data, intelligence, and action in a layer above the source systems. The table outlines the difference for each.
When a CIAH fits: three hard indicators
Not every mid-market business needs a CIAH. Three indicators decide whether the investment pays off: number of source systems used daily, time spent on a recurring weekly question, and whether a concrete action would follow the answer. With all three yes, the use case is real. With one or two, look at single-process workflows first.
- Three or more source systems your team logs into daily. ERP plus CRM plus email plus an Excel workbook plus accounting, the typical mid-market picture.
- At least one weekly question that takes more than 30 minutes to answer today. Examples: which projects are sliding? Which customers have open complaints? Where is inventory running thin?
- If the answer existed, a concrete action would follow. Send the email, place the order, dispatch the technician. Not "interesting to know," but "we would do something with it."
All three yes: a CIAH is probably the highest-impact thing you can buy in the high five-figure range in mid-market. Two yes: consider agentic workflows in individual processes first. One yes: an integration pipeline plus a good dashboard is usually enough.
The anti-SAP pitch
Out of the spreadsheet economy without an SAP-sized invoice. Anyone starting an 18-month S/4HANA project in 2026 should ask honestly whether the business will run better afterwards or just more expensive. A CIAH bets the other way: measurably better in 12 weeks, without touching the system of record.
Discovery sprint: two weeks, agent-assisted business-logic mapping
CIAH licenses are not for sale. Every CIAH is built custom, because the pain points differ by industry and by company. The entry point is a two-week Discovery Sprint that maps the business logic, produces a clickable prototype for the two highest-priority workflows, and ends with an honest build decision. Investment: 5,000 to 10,000 euro. From there it is either a green light for the full build (8 to 12 weeks) or a clear "no, this won't pay off in this shape."
The sprint is agent-assisted. A large part of classical requirements analysis no longer happens in ten-day workshops in 2026; it runs in parallel with AI agents inspecting databases, generating process descriptions, and proposing anomaly rules. That is the lever the agentic mid-market playbook for 2026 is built around.
Frequently Asked Questions
The most common questions in first calls cluster around definition, contrast with ERP, cost, build time, and compatibility with the existing stack. The five answers below summarise the 2026 picture.
What is a Central Intelligence and Action Hub (CIAH)?
A CIAH is a custom software layer that sits on top of existing business systems like ERP, CRM, accounting, and email. It unifies data, detects anomalies, and triggers concrete actions. Unlike an ERP, it replaces nothing. Unlike BI, it doesn't just display data; it acts in the same step.
How is a CIAH different from an ERP like SAP?
SAP S/4HANA is a transactional system handling accounting, logistics, procurement, and Human Resources (HR). A CIAH sits above such systems and uses them as data sources. SAP migrations take 18 to 36 months, cost millions, and fail in 60 percent of cases on budget or schedule. A CIAH project ships in 8 to 12 weeks at 30,000 to 80,000 euro.
How much does a CIAH project cost?
The Discovery Sprint costs 5,000 to 10,000 euro over two weeks. The full build is typically 30,000 to 80,000 euro over 8 to 12 weeks, depending on number of source systems, workflow complexity, and how many role surfaces are needed. Run cost afterwards: low four figures per month for hosting, model usage, and maintenance.
How long does implementation take?
Discovery Sprint: two weeks. Full build: 8 to 12 weeks, with the top three to five workflows live in production. After that, iterative expansion in two- to four-week increments. The answer is deliberately not "we will be back in 18 months."
Do we need to replace our existing SAP or NetSuite?
No, you do not, and that is exactly the point of the CIAH model. The Hub sits above the ERP, not in its place. Existing contracts, processes, data, and licenses stay untouched, including the customizations you have already paid for. A CIAH is compatible with SAP, NetSuite, DATEV, Sage, Microsoft Dynamics, and any of the larger systems running in mid-market today.
Step 1: an honest stocktake
If the three indicators above came back as yes and the next S/4HANA milestone is sitting in your drawer, it is probably time for a second opinion. Better to act before the next mid-market project joins the 60 percent that fail. At IJONIS we build exactly this layer, Discovery Sprint first, full build after, including a clickable prototype and a mapped business logic. The Services page is the entry point for a first sketch and a no-strings conversation.
Conclusion: operational backbone beats another ERP
Key takeaway: The most important investment decision in mid-market in 2026 isn't "which ERP". The real question is whether to build the layer above existing systems, or burn the next six-figure sum on an S/4HANA project. Such projects fail 60 percent of the time.
The bottom line: A CIAH replaces nothing. The Hub uses what is already in place. The Hub unifies data, detects anomalies, and triggers actions. Build time: 8 to 12 weeks. Investment: high five figures. No new ERP risk.
Three yes-indicators, then Discovery Sprint.
- three or more source systems. Daily logins.
- one recurring weekly question. More than 30 minutes to answer today.
- one concrete action. Would follow the answer.
The sprint costs 5,000 to 10,000 euro and produces an honest decision in two weeks.
Operational backbone, not another ERP. That is what the CIAH is built for.
