Zero-spend default
Every AI feature ships switched to a no-cost provider by default. A business only starts paying for AI inference once it deliberately configures a paid provider — nothing runs up a bill silently in the background.
AI ACROSS THE LEDGER
AI that explains your numbers — and never moves them without you.
Across all eleven Enfinity ERP workspaces, AI is built to one design: compute the result deterministically first, add an explanation second, log it to an advisory ledger third, and leave the decision to a named person every time.
Most ERP vendors talk about AI as a black box that decides for you. Finance and operations teams in the GCC do not want a black box near their general ledger, their tax filings or their supplier payments — they want faster clarity on what already happened, without giving up control of what happens next. A finance controller who cannot explain why a system posted a payment or filed a return to their own auditor has not been helped by that system, no matter how sophisticated the model behind it. Enfinity ERP was built around that distinction from the first line of code, not retrofitted with a chatbot after launch, and the design choice shows up the same way in every one of the eleven workspaces described on this page.
How it works
- Compute
The deterministic result — a variance, a match, a score, a forecast — is calculated first from your own posted data, with or without any AI provider configured.
- Explain
If an AI provider is configured, the model adds a plain-language explanation of the deterministic result — it does not recompute or override it.
- Log
Every suggestion, score and flag is written to an advisory ledger with an audit event, so what the system suggested and what a person did are both permanently on record.
- Decide
A named human role — the buyer, the accountant, the credit controller, the treasurer — reviews the suggestion and decides whether to act on it.
- Post
Only a person's confirmation triggers a posting, a payment, a reconciliation, a reminder send or a tax submission — never the AI itself. The advisory ledger entry stays attached to that record permanently.
Capabilities
Metered and budgeted centrally
Token usage is metered against a monthly budget and a per-company daily cap set once at the platform level, not re-configured per module — so a runaway AI job in one workspace cannot blow the budget for the rest of the business.
One advisory ledger, every module
A bill-scan draft, a duplicate-invoice flag, a supplier score and a bank-match suggestion all land in the same underlying advisory ledger, so an auditor or a controller can review every AI touchpoint across ERP from a single place.
Purchase-side AI leads the suite
Bill scan, duplicate-invoice detection, supplier scoring, budget checks, line sanity checks and GL-coding suggestions make Purchase the module with the deepest AI coverage in Enfinity ERP today.
Banking & Cash Flow AI comes second
Bank-match suggestions and cash-flow pressure explanations are the second-deepest AI surface, because reconciliation and liquidity forecasting are where a busy finance team most needs a shortlist instead of a blank screen.
No AI in the customer/supplier portal
The one workspace where an external party signs in has no AI feature at all — an internal advisory ledger entry never becomes a customer- or supplier-facing message.
Deterministic-first variance and impairment reads
Consolidation, fixed-asset and GL variance features flag what already changed in the numbers — the model explains a computed material variance; it never estimates one from scratch.
One design, eleven workspaces
The compute-explain-log-decide sequence is not a Purchase-specific pattern reused elsewhere by coincidence; it is a single base class every AI-assisted feature in Enfinity ERP inherits, so a new AI capability added to any module automatically behaves the same way.
Explanations, not decisions, on tax exceptions
Tax & Compliance's AI reads an already-computed VAT exception and explains why it looks unusual, drawing on the entity's own filing history — it never files, clears an invoice, or asserts compliance on the business's behalf.
A single provider setting, budgeted once
Switching from the zero-cost default to a paid AI provider is a single platform-level configuration change, not eleven separate module settings a business has to find and flip one by one.
Explanations grounded in your own history, not a generic model
A supplier score, a GL-coding suggestion or a collections priority is calculated from the requesting company's own posted transactions and master data — never from a generic industry average or another company's activity — so the explanation a controller reads is always about their own business, not a statistical stranger's.
An audit trail that survives staff turnover
Because every suggestion and every human decision on it live in the same advisory ledger, a new controller inheriting a role can see exactly what the system recommended on a past bill or bank line and what their predecessor actually chose to do — institutional memory that does not depend on one person remembering.
Built to work at zero cost first
Because the deterministic calculation is always the primary result and the AI explanation is always secondary, a small business that never configures a paid provider still gets every score, flag, match and forecast this page describes — it simply reads the number without the accompanying sentence of context.
A pattern any auditor can follow
An external auditor reviewing ERP for the first time does not need to learn eleven different AI behaviours — one explanation of the compute-explain-log-decide sequence covers every advisory feature in every workspace they will encounter during the engagement.
Region-native
Advisories respect the same country boundaries as the ledger
An AI explanation of a VAT exception or a withholding calculation stays scoped to the country profile that produced the number — it never blends rules across a Saudi Arabia entity and a UAE entity in one answer.
Bilingual explanations
Every AI explanation renders in the same language as the screen it appears on, Arabic included, using the finance terminology your team already uses day to day.
No cross-company learning
A supplier score or a GL-coding suggestion is calculated from one company's own dedicated database only — the advisory engine never pools patterns across different companies, so one customer's purchasing behaviour never shapes another customer's suggestions.
Editions
| Small | Standard | Professional |
|---|---|---|
| Zero-spend AI on core modules | Included | Full advisory surface across all eleven workspaces |
Works with
- Enfinity HCM
HCM's own AI-assisted payroll anomaly checks share the same advisory-ledger and zero-spend-default design as ERP's AI, so a group standardises on one AI governance model.
- Enfinity CRM
CRM's own AI features draw on the same Platform.AI gateway, so a lead-scoring explanation in CRM and a supplier-scoring explanation in ERP are governed the same way.
- Enfinity Restaurant
Restaurant's AI-assisted recipe and stock features post through the same Inventory and Accounting workspaces ERP's own AI advisories touch, keeping one advisory ledger across both.
The Enfinity platform
- The AI gateway
Every AI feature described here is a consumer of the shared platform gateway — never a direct call to a model provider from ERP's own code.
- Reporting & personalization
The advisory ledger itself is queryable through the same reporting engine as any other ERP data, so a controller can build a custom view of AI activity without a developer.
FAQ
Does Enfinity ERP need an AI provider to function?
No. Every deterministic calculation — variance, match, score, forecast — runs identically with or without a provider configured; only the plain-language explanation is missing without one.
Can AI post a transaction on its own?
No, anywhere in Enfinity ERP. Every AI feature in this suite is advisory; a named human role must act for anything to post, pay, reconcile or send.
How much does the AI cost to run?
Nothing by default — every feature starts on a zero-cost provider. Costs only begin once a business deliberately configures a paid provider and stay within the budget it sets.
Which module has the deepest AI coverage?
Purchase, with six AI-assisted features spanning bill scan to GL-coding suggestions, followed by Banking & Cash Flow.
Is every AI suggestion recorded?
Yes — every suggestion, score and flag writes to an advisory ledger with an audit event, whether or not a person acted on it.
Does the portal for customers and suppliers use AI?
No — it is the one workspace in Enfinity ERP with no AI feature at all, by design.
Can a business turn off AI for one module but keep it for another?
The provider setting is platform-wide, but every feature already degrades gracefully to its deterministic result with no explanation when no provider is configured, so the practical effect is the same as an off switch.
Who can see what the AI suggested versus what actually happened?
The advisory ledger keeps both records side by side, so a controller or auditor can compare a suggestion against the actual posted, paid or sent outcome at any time.
Does AI ever contact a customer or supplier directly?
No. It prepares a reminder draft or a summary for a human to review; every message that actually reaches a customer or supplier is sent by a person.
Will a new AI feature added later follow the same rules?
Yes — because every AI-assisted feature inherits the same base advisory pattern, a future capability added to any of the eleven workspaces starts from the same deterministic-first, human-reviewed design described on this page rather than a one-off exception.
Where can I read the full list of AI features per module?
Each of the eleven workspace pages lists its own AI-assisted features with the same boundary language used here, alongside the deterministic capabilities and region-native features that surround them.