Exact Online + private AI
AI for Exact Online, private and grounded in your own data
Short answer
Exact Online is true multi-tenant SaaS, so 'on-prem AI for Exact Online' really means a private model beside it, grounded on data pulled through Exact's REST API into a tenancy you control, not inside Exact's own cloud. Done that way, finance and operations teams get plain-language answers on cash position, AR aging, and multi-division consolidation without waiting on a Power BI report or hitting Exact's default reporting.
- ERP
- Exact Online
- Industries
- Wholesale Distribution, Light Manufacturing, Professional Services
- Written for
- Finance Director
Exact Online is popular with small and mid-size finance and operations teams precisely because it is simple, cloud-native accounting and ERP with a clean interface. That simplicity has a flip side: reporting beyond what Exact's built-in dashboards offer usually means exporting to Excel or building a Power BI report against the API, and finance teams without a dedicated analyst do that work themselves.
The questions finance directors actually ask, what's our cash position after these open invoices clear, which customers are drifting into 60-plus days past due, how does this month compare across our different country divisions, are all answerable from Exact Online's data, but not in plain language today. Every answer requires opening the right screen or building the right report first.
Manufacturing and wholesale customers on Exact Online Premium add another layer: stock, BOM, and order data that finance rarely touches directly, so questions that span both financials and operations, like the margin impact of a component price change, require pulling data from two places and reconciling it by hand.
Because Exact Online has no on-prem deployment option, adding private AI here means something specific: pulling data through Exact's REST API into infrastructure you control, and running the model there. That is a real and useful architecture, it just needs to be understood clearly for what it is, data leaving Exact's tenant into your own, under your own governance, not AI running inside Exact's cloud.
What usually gets in the way
The problems we hear most from finance director teams running Exact Online.
Reporting beyond Exact's dashboards means a manual export
Anything outside Exact Online's built-in reports usually becomes an Excel export or a Power BI build, repeated whenever the question changes.
Cash flow and AR aging questions require opening the right screen
Finance directors ask plain questions about cash position and aging receivables that Exact Online can answer, but only through its standard reporting path, not a quick query.
Multi-division, multi-country administrations don't roll up on their own
Each Exact Online 'division' is effectively a separate environment, so combining numbers across countries or entities is a manual consolidation exercise.
Manufacturing/stock questions and financial questions live apart
Exact Online Premium customers with stock and BOM data still have to reconcile operational and financial data by hand to answer margin or cost-impact questions.
API rate limits and OAuth token cycles trip up naive integrations
Exact Online's REST API enforces per-minute rate limits and OAuth2 token refresh, which catches out integrations that weren't designed with those constraints in mind from the start.
Where AI earns its place in Exact Online
Each use case names the ERP objects it reads or writes, so your ERP team can judge the integration effort before anyone commits budget.
Natural-language finance queries
Finance staff ask directly for cash position, open invoice totals, or budget-vs-actual by cost center, without opening a Financials report.
Touches: GL, AP, AR transaction and balance data via the Exact Online Financials API
Outcome: Answers in seconds instead of building or re-running a Power BI report
Cash flow and AR aging copilot
Summarizes projected cash position from open AR and AP, and flags customers drifting into higher aging buckets.
Touches: Accounts receivable aging, accounts payable due dates, bank/cash balances
Outcome: Earlier visibility into collection risk and short-term cash pressure
Manufacturing stock and BOM assistant
For Exact Online Premium manufacturing/wholesale customers, answers stock level, BOM, and reorder questions and ties them to margin impact.
Touches: Item and stock records, bill of materials, sales order and purchase pricing
Outcome: Finance and operations get a shared answer instead of reconciling two exports by hand
Multi-division consolidation agent
Queries across separate Exact Online divisions (countries or entities) and consolidates the answer into one response.
Touches: Financial data across multiple Exact Online division/API scopes
Outcome: Same-day multi-entity answers instead of a manual monthly consolidation
Invoice and document AI feeding AP
Extracts data from incoming vendor invoices and matches it against POs and receipts before posting into Exact Online.
Touches: AP invoice documents, purchase order and goods-receipt data
Outcome: Fewer manual keying errors and faster invoice approval cycles
Budget vs. actual variance narrative
Generates a plain-language explanation of budget-to-actual variance by department or cost center for management reporting.
Touches: Budget records, GL actuals by cost center
Outcome: Management gets a written narrative alongside the numbers without finance drafting it manually every month
CRM-to-financials handoff assistant
Answers questions that span the quote-to-order-to-invoice cycle, flagging where a deal in CRM hasn't converted to an invoice on schedule.
Touches: Exact Online CRM opportunity and quote data, sales order and invoice records
Outcome: Sales and finance catch stalled deals or billing gaps earlier
Reference architecture
Because Exact Online has no on-prem option, data is pulled through its REST API into infrastructure the customer controls; the model, retrieval layer, and governance all run there, beside Exact Online rather than inside it.
- 1
ERP connectors
OAuth2-authenticated access to the Exact Online REST API, scoped per division, with sync scheduling designed around API rate limits.
- 2
Data and semantic layer
Maps Exact Online's Financials, CRM, and (where applicable) manufacturing data into plain business vocabulary.
- 3
Model serving
An open-weight model served in a private, single-tenant environment the customer controls, separate from Exact's own cloud.
- 4
Retrieval and agents
Text-to-SQL/API-query over synced financial and operational data, plus agent tools for drafting narratives and invoice matching.
- 5
Governance and audit
Access mapped to Exact Online roles where possible, with full logging of what data was pulled and what the model was asked.
Integration notes for your ERP team
- Exact Online's REST API uses OAuth2 with token refresh and per-minute rate limits; sync jobs need to be designed around those limits from the start, not bolted on after hitting them.
- Each Exact Online 'division' (typically a country or legal entity) is a separate API scope, so multi-entity consolidation requires an explicit federation layer, not a single query.
- Exact Online supports webhook-style change notifications for some entities, which can reduce polling load compared to a naive full-sync approach.
- Write-back, such as posting a journal entry or updating a CRM record, should go through the same API validation Exact's own UI relies on, with human approval before anything posts.
- Exact continues to expand its own built-in reporting and automation features, so scope custom AI use cases to genuinely under-served questions rather than duplicating what Exact Online already surfaces.
- Because Exact Online has no on-prem deployment, be explicit in any GDPR or DPA conversation that private AI means synced data leaving Exact's tenant into the customer's own environment.
Deployment options
Private / sovereign cloud (EU region)
Companies wanting data residency alignment with Exact Online's own EU hosting, without owning hardware
Data synced via API into an EU-region private tenancy; the model runs there under the customer's own governance.
On-prem inference with cloud-synced data
Companies that want the model itself physically inside their own network even though the source ERP is SaaS
A scheduled sync pulls Exact Online data into a local cache; the model runs entirely on customer hardware from that cache.
Hybrid
Multi-country groups needing both local-language handling and a central consolidated view
Regional private inference for local questions, with a central private tier for cross-division consolidation.
Compliance and data control
How the architecture supports your obligations. Certification and accountability stay with your organisation; the design keeps the evidence straightforward.
GDPR
Pulling data via API into a customer-controlled private tenancy, with a clear data processing agreement, keeps the AI layer inside the same GDPR obligations Exact Online's own EU hosting already respects.
EU AI Act
Deploying the model privately with logged queries and human review of generated content supports the documentation and transparency expectations that apply to AI systems used in business decision-making.
Accountant/auditor data handling
Because many Exact Online customers work through an external accounting firm, access controls and audit logs are designed to respect that existing relationship rather than bypass it.
Data residency for multi-country groups
EU-region private hosting keeps synced financial data within the same broad jurisdiction as Exact's own hosting for customers with residency requirements.
Where Netray fits
ERPray
Exact Online is not a listed connector today; natural-language question-answering over synced financial data is built on ERPray's ERP-agnostic connector architecture.
DataRay
A good fit where Exact Online data needs to be combined with other systems, such as a separate WMS, CRM add-on, or document store, in one queryable layer.
Custom build
Multi-division consolidation and invoice-matching agents are typically specific enough to a customer's Exact Online setup to warrant purpose-built connectors.
How an engagement runs
Phase 1 . 2 weeks
Discovery
- -Exact Online API scope and division review
- -data residency requirements confirmed
- -prioritized use-case list
- -hosting decision
Phase 2 . 6 weeks
Pilot
- -private model and sync pipeline stood up
- -connector to one Exact Online division
- -2-3 use cases live (finance query, AR aging)
- -finance team feedback loop
Phase 3 . 3-4 weeks
Production
- -governance and audit logging hardened
- -budget variance and invoice-matching use cases rolled out
- -role mapping to Exact Online permissions
- -DPA/GDPR documentation finalized
Phase 4 . ongoing
Scale
- -additional divisions/countries connected
- -multi-division consolidation agent
- -manufacturing stock/BOM use case if on Premium
- -quarterly use-case review
Questions to ask any vendor, including us
A short list that separates real Exact Online AI work from a chatbot demo.
- Since Exact Online has no on-prem option, exactly where does our synced data live once it leaves Exact's tenant?
- Does the AI ever write back into Exact Online, or is it strictly read-only?
- How is the API sync designed around Exact Online's rate limits and OAuth token refresh?
- How do you handle multiple Exact Online divisions if we operate in more than one country?
- What does the data processing agreement say about this synced data specifically?
- Who audits what was pulled from Exact Online and what the model was asked?
- How does this differ from Exact Online's own built-in reporting and automation features?
Frequently asked questions
Can Exact Online run AI fully on-premises?
Not literally inside Exact Online itself, since it is multi-tenant SaaS with no on-prem deployment option. A private AI layer works by pulling data through Exact's REST API into infrastructure the customer controls, where the model then runs privately.
What is the fastest AI win for an Exact Online finance team?
Natural-language queries over cash position and AR aging are usually fastest to stand up, since that data is already well structured in Exact Online's Financials API. Most teams see a working pilot within about six weeks.
Does Exact Online already have built-in AI features?
Exact has been adding automation and insight features to Exact Online over time. Whether they cover a given question depends on the release and module; a private AI layer is typically scoped to fill gaps rather than duplicate what Exact Online already surfaces.
Is pulling data out of Exact Online for AI GDPR-compliant?
It can be, when governed by a clear data processing agreement, EU-region hosting for the private AI tenancy, and logged, access-controlled use of the synced data. This is a standard pattern for private AI beside any EU-hosted SaaS ERP.
Can this handle multiple countries or divisions in Exact Online?
Yes, but it requires an explicit consolidation layer, since each Exact Online division is a separate API scope. The AI queries each division and combines the results, rather than treating multi-entity data as a single source automatically.
Does this work for Exact Online Premium manufacturing customers?
Yes. Stock, BOM, and order data available through Exact Online Premium's manufacturing features can be indexed alongside financial data, which is particularly useful for questions that span operations and margin together.
What does it cost to add private AI beside Exact Online?
For a single-division pilot covering two or three use cases, expect a cost in line with a focused mid-size integration project. Multi-division consolidation and manufacturing use cases add scope and cost from there.
Related guides
Natural Language Query for ERP Data: Ask SAP, Infor, or Oracle a Question in Plain English
See how natural language query over SAP, Infor, Oracle, and NetSuite data works: grounded text-to-SQL, role-based permissions, and a visible audit trail.
RAG + SQL + permissionsA Private LLM Grounded on Your ERP Data
How a private LLM answers questions on your ERP data: RAG plus text-to-SQL, role-based permissions inherited from the ERP, and where each fits.
Agents + approval gatesAI Agents for ERP, Running On-Prem
A practical guide to on-prem AI agents for ERP: what they can safely automate, where human approval belongs, and how to design the guardrails.
ERP AI Buyer GuideHow to Choose an ERP AI Implementation Partner
A CIO checklist for picking an ERP AI implementation partner: the architecture questions to ask, red flags, pricing models, and what to demand in the SOW.
ERP AI Cost GuideWhat ERP AI Actually Costs: A CFO's Guide
A CFO's guide to what ERP AI actually costs: GPU hardware, model licensing, integration and connector work, and realistic ongoing run-rate ranges.
QAD + on-prem AIAI for QAD Adaptive ERP in automotive and industrial manufacturing
Add AI to QAD Adaptive ERP or Enterprise Edition for automotive and industrial manufacturing, grounded on QXtend and QAD's API layer, on-prem or private cloud.
Plan it with numbers
Cloud vs On-Premise ERP Cost Calculator
Put cloud subscription and on-premise ERP costs side by side over 5 years, including the maintenance, infrastructure, and staffing lines that skew the comparison.
Free ToolOn-Prem vs Cloud AI Cost Comparison Calculator
Compare your projected cloud LLM API bill against an amortized on-prem GPU deployment and find the breakeven month.
Free ToolERP AI Maturity Assessment
Benchmark how deeply AI and automation are embedded in your ERP operations, from data foundations to autonomous agents, across four maturity levels.
GuideERP Cloud Migration Strategy Guide
Plan your ERP cloud migration with a proven strategy covering assessment, architecture selection, data migration, and go-live. Includes AWS, Azure, and GCP options.
GuideEU AI Act Implications for On-Prem AI Deployments
EU AI Act implications for on-prem deployments: risk tiers, high-risk obligations, and how self-hosted models simplify enterprise compliance.
GuideERP Copilots: Measuring Real User Productivity Gains
ERP copilots promise productivity, but what do users actually gain? Measured results, metrics that matter, and how to deploy copilots for SyteLine and LN.
Talk it through with an engineer who knows Exact Online
Bring one real question your team cannot answer from the ERP today. We will map the data path, the model, and where it runs, and tell you honestly if AI is the wrong tool for it.