A category of enterprise resource planning vendor that barely existed two years ago now has funded companies, reference customers and a coherent pitch. The AI-native challengers argue that the incumbents have spent the last eighteen months adding assistants to software designed in the nineties, and that the correct response is to rebuild the system around the model rather than beside it. It is a good argument. It is also the argument every enterprise software challenger has made since client-server, and the question worth asking at the start of this year is which parts of it are different this time.
An assistant bolted onto a transactional system helps you use the screens faster. Rebuilding around the model means having fewer screens — and that is a much larger claim than the demonstrations show
Here is how to evaluate the category without either dismissing it or buying the narrative.
What the challengers are actually building
Three things, in varying combinations. Conversational entry in place of forms, where a purchase requisition or a journal entry is described rather than keyed. This demonstrates extremely well and is the easiest part to replicate, which is why the incumbents already have it. Document-first processing, where an invoice, contract or order confirmation is the input and the structured record is the output rather than the other way round. This is genuinely better architecture for the messy end of finance and operations. Agentic execution, where the system proposes and then carries out multi-step work — chasing an approval, matching a receipt, raising a credit note. This is where the value is and where the evidence is thinnest.
Where the incumbents are actually vulnerable
Not in features. In the parts that are expensive to retrofit. Data model rigidity. Legacy systems encode assumptions about how a transaction must be shaped, and a model-first product can accept ambiguity and resolve it later. That is a real architectural advantage and hard to copy. Implementation economics. If configuration genuinely happens through description rather than consulting days, the cost structure of the whole market changes — and the incumbents' partner channels have no interest in that happening. The long tail. Thousands of companies run a core system plus fourteen spreadsheets because the system could not accommodate their process. That tail is where a flexible product wins first.
Where the challengers are weak, and it is the same place every time
Statutory and regulatory coverage. Tax regimes, e-invoicing mandates, statutory reporting formats, audit requirements and local payroll rules are enormous, unglamorous and non-negotiable, and they are where new entrants discover that the boring eighty per cent of an enterprise resource planning system is boring for a reason. Also: multi-currency and consolidation at depth, the audit trail an external auditor will accept, the integration ecosystem, and the simple question of whether the company will exist in seven years. That last one is not cynicism; it is the single most common reason a capable challenger loses a deal it should win.
How to run the evaluation
Do not run a feature comparison, because you will lose. Run your three ugliest real processes through both and look at what happens at the exceptions, not the happy path. Ask the challenger to show you its statutory coverage for every country you operate in, by name. Ask both for the audit evidence an external auditor receives. And price the incumbent's assistant modules honestly, because the comparison people actually make — new product against legacy system as configured in 2016 — is not the comparison they will be living with. For most established mid-market organisations the defensible answer this year is a serious pilot in one domain rather than a core replacement. For a company implementing its first real system, the calculus is genuinely different.
| Decision area | Evidence to request |
|---|---|
| Process handling | Run difficult real processes and inspect exceptions. |
| Statutory coverage | Written country-by-country coverage and named local references. |
| Audit trail | The evidence your external auditor would receive. |
| Commercial comparison | Include incumbent assistant modules in the price. |
| Continuity and exit | Assess survivability, local support and data export. |
Qualitative summary of this article's source text, not a measured outcome or performance estimate.
Practical Guidance for Emerging ERP Landscape Briefing
- Test your exceptions, not the demonstration workflow.
- Demand country-by-country statutory coverage in writing.
- Ask what an external auditor receives as evidence.
- Check e-invoicing mandate support for every market you invoice in.
- Price the incumbent's assistant modules into the comparison.
- Assess vendor survivability as a formal criterion, not a worry.
- Consider a domain pilot before a core decision.
- Keep your data exportable whichever way you go.
The Regional Angle
The first filter here is statutory, and it eliminates more challengers than anything else. A system used in this region needs Saudi e-invoicing integration working against the tax authority's actual platform, value added tax treatment across several Gulf jurisdictions with different rules, United Arab Emirates corporate tax now in its second cycle with free zone qualifying income to compute, Arabic-language statutory documents, and wage protection system files in the format each country's ministry expects. None of that is interesting and all of it is mandatory. Ask for named live customers in your specific country rather than regional coverage claims, because "we support the Gulf" frequently means one implementation in one emirate. The second is a genuine opening for the challengers that regional buyers should take seriously. A very large number of family groups and mid-market companies here run diversified portfolios — trading, contracting, real estate, retail and services under one ownership — where no single industry template fits and the incumbent's answer has always been an expensive customisation programme per entity. A system that adapts to described process rather than requiring configuration against a fixed model is exactly suited to that shape, and the cost of the alternative is well understood by anyone who has sat through a group implementation. This is the strongest local use case for the new category, and it is worth piloting in the least regulated entity in the group. The third is about how these decisions actually get made in regional groups, and it cuts against the challengers in a way their sales teams underestimate. The decision-maker is frequently the owner or a family board rather than a technology committee, the relationship with the existing vendor or implementation partner often predates the current management, and continuity is weighted heavily against novelty. A pilot that succeeds technically will still lose if nobody can answer the question of who supports it in five years and whether there is a partner in-country who can be reached on a Thursday afternoon. Challengers entering this market win on local presence before they win on architecture, and buyers should treat the absence of a local support entity as a substantive objection rather than a detail to resolve later.
The objection worth taking seriously
The strongest objection is historical. Every generation of enterprise software has produced challengers who argued that the incumbents were structurally obsolete, and the overwhelming majority were acquired, pivoted or quietly disappeared while the incumbents absorbed the interesting ideas and kept the customers. Enterprise resource planning is especially resistant because switching costs are enormous, the regulatory surface is vast, and the buyer's downside from a failed migration dwarfs the upside from a better interface. On the base rates, the rational prediction is that the incumbents ship comparable capability within two years and the category consolidates. That is probably right about most of these companies, and a buyer betting the core system on one of them is taking a risk disproportionate to the gain. What the base-rate argument misses is that the incumbents' constraint this time is not engineering, it is the implementation channel. The assistants shipping from the large vendors are good, and they are being sold through partner networks whose revenue depends on configuration days. A product that genuinely reduces implementation effort is a threat to the people who sell it, which is why incumbent assistant features consistently make consultants faster rather than making consultants unnecessary. If the challengers win anything durable it will be there — in the economics of getting live — rather than in the software itself. Watch implementation cost, not feature parity, because that is the number that will tell you whether this generation is different.
Common Questions
Should we delay a planned implementation to see how this develops?
Rarely. A year of deferral costs more than the difference between a good system chosen now and a better one chosen later, and the category will not settle within your decision window.
Can these products handle statutory reporting?
Some can, in some countries. Verify per jurisdiction with named references, because this is where optimistic roadmaps concentrate.
Is conversational entry actually useful?
For occasional and field users, meaningfully so. For a clerk processing two hundred lines a day, a well-designed form is still faster, and any vendor claiming otherwise has not watched one work.
What should we expect over the next twelve months?
Expect the incumbents to ship agentic features aggressively and price them as upsell. Expect at least one prominent challenger to be acquired. Expect the first honest public accounts of AI-native implementations that ran long, which will be useful reading. And expect the durable differentiator to turn out to be implementation cost rather than model capability.
Emerging ERP Landscape Briefing — we test the challengers against your statutory requirements and your ugliest process, not against a demonstration.