The board wants to know what your AI plan is. Your team, meanwhile, is still entering supplier invoices by hand and chasing reconciliations at day eight of the close. AI in finance becomes useful when it removes a specific task or helps someone spot a problem earlier. The first question is where your team loses time each month.
Gartner’s survey of 183 CFOs and senior finance leaders in May and June 2025 found that 59% reported using AI in their finance function. That figure measures adoption without establishing how much work the technology saves. For a UK Finance Director, the next questions are practical: what can your team use now, what does it need to work, and who checks the result?
What Does AI in Finance Actually Do Today?
AI in finance can help your team retrieve information, process supplier invoices, and identify unusual transactions. Gartner’s 2025 survey found that knowledge management was the most common use among respondents already using AI, at 49%, followed by accounts payable process automation at 37% and error and anomaly detection at 34%. Knowledge management covers organising and retrieving information; it extends beyond querying ledger balances. These applications support familiar work, from finding an answer in company records to checking an invoice before posting. Their usefulness depends on the task and the quality of the review.
How Are UK Businesses Using AI in 2026?
AI use is growing across UK businesses, although adoption remains limited in depth. The Office for National Statistics (ONS) reported that around 35% of businesses with 10 or more employees used at least one AI technology in June 2026, compared with around 12% in September 2023. The average number of AI technologies per adopting business rose from about 1.4 to 1.6. Those figures cover businesses across sectors, rather than finance departments specifically. For your team, a useful starting point is to identify one repeated task, measure the time it takes, and test whether AI reduces that work without increasing corrections.
Can You Trust AI Output in Your Finance System?
Human review and good data make AI output easier to assess, but neither guarantees accuracy. You need to understand the source records, the permissions the feature uses, and what happens after it produces an answer or flags a transaction. A useful check is whether your reviewer can explain the result and trace it back to the underlying evidence.
Who reviews the output
Give each AI-supported process a named reviewer and define what they must check. For an invoice, that includes the supplier, amount, coding, and supporting document. For an answer about spending, check the reporting period, entities, and included transactions. Decide which exceptions need escalation and how the reviewer records the action taken. That gives your team a repeatable review process. Our guide to what auditors look for in your finance system covers the wider control requirements.
Why data quality comes first
Gartner identifies data literacy, technical skills, and inadequate data quality or availability as major obstacles to AI adoption in finance. Review those foundations before adding another tool. Duplicate supplier records, inconsistent department codes, or missing transactions make the underlying data harder to interpret. Agree who owns those records and how your team corrects them. When you test an AI feature, compare its output with work your team has already checked, including exceptions. Record errors and extra review time alongside any time saved. That supports a decision about whether to expand its use.
How Does Sage Intacct Use AI in Accounts Payable?
Sage Intacct uses AI to turn supplier documents into draft bills, match invoice lines to Purchasing transactions, and flag unusual invoices before posting. Sage’s documentation lists these capabilities as available in the UK, with the relevant subscriptions and configuration. Each supports a different part of the AP process.
Draft bills from supplier documents
AP Automation reads supplier documents that you email or upload and creates draft bills for review. It requires an AP Automation subscription alongside Accounts Payable. Your reviewer checks the proposed details and corrects them where needed before processing the bill. That reduces repetitive data entry while keeping the AP team’s judgement in the workflow.
Line-level purchase order matching
Sage’s 2026 R1 release notes confirm general availability of line-level matching in all regions. It matches supplier invoice lines to the source Purchasing transaction and flags price or quantity differences. It supports three-way matching between purchase orders, receipts, and invoices, with tolerance thresholds for minor differences. The documented requirements include Purchasing, Accounts Payable, AP Automation, and Sage Cloud Service. For a construction or distribution team using purchase orders, the reviewer can investigate the flagged lines within the configured approval process.
Anomaly detection before posting
Sage added AP anomaly detection in 2026 R3. It flags unusual amounts, unrecognised supplier email addresses, and transactions with several anomalies. Reviewers see the indicators in the Automated transactions list once an administrator adds the Anomaly column to a custom view. It requires Accounts Payable and AP Automation subscriptions and is available in all regions. Our round-up of what’s new in Sage Intacct 2026 R3 covers the wider release.
An anomaly flag prompts a review; it doesn’t stop the transaction from posting.
Can AI Speed Up the Month-End Close in Sage Intacct?
Sage Intacct combines AI checks with workflow automation to support a faster close. Sage’s release notes confirm that Close Automation has been generally available in the UK since 2025 R3, with a separate Close Automation subscription. It helps your team track tasks, compare balances, and review budget variances. The scope matters: tracking a checklist is workflow automation, while detecting an unusual posting uses machine learning.
Close Workspace and Close Assistant
Close Workspace provides shared checklists and a progress monitor across entities and assignees. Sage made it generally available in the UK in 2025 R4 as part of Close Automation. Close Assistant shows standard period-end tasks across core applications. That gives managers a central view of outstanding work. Close Analytics adds days-to-close and bottleneck reporting for existing Early Adopter participants; Sage’s documentation says new registration for that programme is closed.
Subledger reconciliation and variance analysis
Subledger Reconciliation Assistant automates some of the comparisons between subledgers and the general ledger. Your team still investigates and resolves differences. Variance Analysis alerts budget owners and finance leaders when actual spending exceeds budget. Sage’s 2026 R1 notes added support for companies with multiple base currencies: reconciliation views show entities sharing a base currency, and variance reviews use entity groups whose members share one. Our guide to automating month-end close with Sage Intacct covers the wider tools.
GL Outlier Detection at approval
Sage includes GL Outlier Detection in the Sage Intacct subscription. It learns historical transaction patterns and flags journal entries outside those patterns during approval. Outlier Assistant can return a flagged entry to its submitter for review before it reaches the approver. You need historical data, configured journal approvals, and the feature enabled. The flag supports the review; the approver remains responsible for deciding whether to approve, correct, or decline the entry.
What Are Sage Copilot and the Finance Intelligence Agent?
Sage describes Copilot as the interface connecting users to its AI agents and features. The Finance Intelligence Agent is one of those agents: it answers natural-language questions about Sage Intacct data. Access depends on your subscriptions and permissions. Sage also includes Search Help with Copilot in the Sage Intacct subscription, which searches product help rather than your financial records.
The Finance Intelligence Agent in the UK
Sage gives the example of asking the Finance Intelligence Agent for the five suppliers with overdue bills and the total outstanding for each. A “How Copilot worked this out” tab explains how it derived the answer. Users need permission to access the data they ask about. The 2026 R3 release notes describe a phased Early Adopter rollout covering the UK, in US English only. Sage notifies eligible companies when their administrator can enable it. Check the answer’s scope and supporting records before using it for a spending decision.
Partner agents and the AI Trust Label
Sage’s 2026 R3 notes introduce a UK-available marketplace for agents from certified partners. Access requires a Finance Intelligence Agent subscription and role-based permissions; purchasing agents also requires a paid AWS account. General marketplace availability therefore doesn’t mean every company can use an agent immediately. Separately, Sage’s AI Trust Label has been live for UK Sage Intacct customers since November 2025. It explains matters such as customer data use and accuracy monitoring. That provides information for your team to assess alongside its own controls.
Which Sage Intacct AI and Automation Features Are Available in the UK?
Sage’s documentation distinguishes general availability from Early Adopter access. The table covers the AI and automation features discussed here, plus AI-powered imports from the 2026 R1 release notes. General availability means a feature is available to eligible customers; subscriptions, permissions, and setup still apply. Statuses checked on 24 September 2026:
| Feature | What it does | UK status |
|---|---|---|
| AP Automation | Creates draft bills from supplier documents | Generally available (AP Automation subscription) |
| Line-level matching | Matches invoice lines to Purchasing transactions and flags differences | Generally available; Purchasing, Accounts Payable, AP Automation, and Sage Cloud Service required |
| AP anomaly detection | Flags unusual amounts and unrecognised supplier emails | Generally available; Accounts Payable and AP Automation required |
| GL Outlier Detection | Flags journal entries outside historical patterns during approval | Included; enable the feature and configure journal approvals |
| Close Automation (Close Assistant, Subledger Reconciliation Assistant, Variance Analysis) | Tracks tasks, assists balance comparisons, and flags budget variances | Generally available; Close Automation subscription required |
| Close Workspace | Shared close checklists and progress monitoring | Generally available; Close Automation subscription required |
| Close Analytics | Days-to-close and bottleneck reporting | Existing Early Adopter participants; new programme registration closed |
| Sage Copilot | Interface for subscribed agents and features; help search is also available | Access depends on feature subscription and permissions; help search is included |
| Finance Intelligence Agent | Answers natural-language questions about permitted financial data | Phased Early Adopter rollout; Finance Intelligence Agent subscription and enablement required |
| Partner agent marketplace | Certified partner agents extend Sage Copilot | UK available; Finance Intelligence Agent access, Marketplace Agents, role-based permissions, and paid AWS account required |
| AI-powered imports | Helps transform and map supported import data, with a preview | Available in the UK; access varies by import type and permissions |
Choose Your First AI Finance Task
Choose the first feature around the work you need to improve. If invoice entry takes most of your AP team’s time, assess AP Automation. If managers chase close updates across spreadsheets, assess Close Workspace. A team already using AP Automation can add the Anomaly column and agree how reviewers investigate flags.
Check the required subscriptions and assign an owner before enabling a feature. Run it through a defined period of work, then compare processing time, corrections, and unresolved exceptions with your starting position. We can help you assess the Sage Intacct configuration and review process your team needs.
If you’d like help choosing a suitable first use case, booking a discovery session is a good place to start.
References
- Gartner: AI adoption in finance, 2025
- ONS: AI in UK businesses, 2023–2026
- Sage Intacct Help: AI features in Sage Intacct
- Sage Intacct Release Notes: Line-level invoice matching
- Sage Intacct Release Notes: AP anomaly detection
- Sage Intacct Release Notes: Close Automation
- Sage Intacct Release Notes: Close Workspace
- Sage Intacct Release Notes: Close Analytics
- Sage Intacct Release Notes: Close Automation with multiple base currencies
- Sage Intacct Help: GL Outlier Detection
- Sage Intacct Help: Sage Copilot
- Sage Intacct Release Notes: Finance Intelligence Agent: features
- Sage Intacct Release Notes: Finance Intelligence Agent: phased rollout
- Sage Intacct Release Notes: AI agent marketplace
- Sage UK: AI Trust Label
- Sage Intacct Release Notes: AI-powered import tools
