Working AI deployed in 4–6 weeks, on the platforms you already use. Liorant builds and manages the AI systems that remove finance's administrative load — for a two-person accounting function and a multi-entity finance department alike.
No open-ended consulting. No transformation programme that ends with a slide deck. Three structured phases — each producing a concrete deliverable.
Free, 30–60 minutes. We map your workflows, confirm which platforms you already licence, and identify the highest-value place to begin. No presentation, no pitch.
4–6 weeks, fixed scope and price. We build the AI system on your platform, test it against real transaction and ledger data, and hand it over with documentation and governance guardrails. A working system — not a prototype.
Monthly, optional. We monitor, optimise, and expand your AI portfolio — and each month you receive a business outcome report written as a leadership narrative, not a technical dashboard.
The administrative load is structural. It runs in parallel to every close cycle, consumes your most experienced people, and scales with transaction volume — whether finance is one person doing everything or a department split across AP, controlling, and FP&A.
Accounts payable staff key supplier invoice data from PDF and email attachments into the accounting system by hand — 10 to 20 minutes per invoice, on every invoice that arrives.
Staff accountants reconcile bank statements against the ledger line by line every cycle, worse across multiple accounts or currencies.
Financial controllers pull data from multiple systems, draft the same recurring journal entries, and track the close checklist by hand — often 5 to 10 business days, with last-minute fire drills.
FP&A analysts compare actuals to budget in a manual spreadsheet exercise each month, with no systematic way to flag which cost centers moved and why.
Compliance and tax specialists compile and cross-check records for filings and audits by hand, under deadline pressure, across every system that holds a transaction.
None of this work requires human judgment. Liorant builds the AI systems that handle it — structured, auditable, and governed to your company standards, regardless of headcount.
Each starts as a single, scoped deployment with a defined deliverable. All connect into coordinated systems as the portfolio grows.
Reads invoices and matches POs in minutes, not days.
An AI capture assistant, built on your platform, extracts vendor, amount, PO reference, and line items into a structured review sheet and flags mismatches before payment.
At scale · automatic posting, payment-batch prep, and duplicate-invoice anomaly detection.
Drafts collection reminders, matched to payment history.
An AI assistant connected to the AR aging export produces a prioritized follow-up list and drafts reminder messages matched to each customer — gentle for reliable payers, firm for repeat late payers.
At scale · auto-triggered reminders and a payment-risk score for likely bad debt.
Pre-matches transactions — your team reviews what's left.
A reconciliation workflow ingests the bank export and the ledger export, auto-matches by amount, date, and reference, and produces a flagged-exceptions list for review.
At scale · multi-entity, multi-currency, and auto-categorized recurring transactions.
Checks every expense line against policy automatically.
An AI reviewer reads submitted receipts and reports, flags policy violations — limits, categories, missing receipts — and produces an approval-ready summary for the finance manager.
At scale · direct intake from the expense platform and real-time submission-time flagging.
Drafts recurring entries and tracks the close checklist.
A close-checklist assistant drafts routine recurring entries — accruals, depreciation, prepayments — for accountant review and tracks which close tasks are pending versus complete.
At scale · variance flagging and an automated close-status view for the finance lead.
Turns the P&L export into a same-day leadership narrative.
An AI assistant connected to the monthly P&L/KPI export generates a plain-language narrative and a one-page KPI summary, ready for the leadership meeting.
At scale · live Power BI / Looker Studio connections and board-deck-ready summaries.
Flags and explains budget deviations by cost center.
A variance workflow compares actuals to budget by cost center, flags variances above a defined threshold, and drafts a plain-language explanation request for the department owner to confirm.
At scale · a rolling forecast updated monthly from recent actuals, plus trend detection.
Compiles and cross-checks records ahead of every filing deadline.
A compliance workflow pulls relevant transaction records against a filing checklist, flags missing documentation or inconsistencies, and produces an organized packet for the accountant or external advisor to finalize.
At scale · ongoing compliance monitoring ahead of each filing cycle.
Each starts as one scoped deployment — then connects into a coordinated operating layer as your portfolio grows.
Start with a use caseMost institutional knowledge sits in file servers, email archives, and individual spreadsheets. Liorant makes it retrievable in plain language, answered from your own files — not from the internet, not from a generic base.
Approval limits, expense policy, and close-checklist templates — answered in plain language, sourced from your internal documentation.
Prior-period entries, recurring accrual templates, and variance explanations — retrievable by account, cost center, or period.
GL mapping, tax codes, and filing requirements (SII, Verifactu, and equivalents) retrieved by natural-language query.
New team members query your real close procedures from day one — no senior accountant required — and the system flags where human review is needed.
Built on the same platforms as your operational AI, and connected to systems already deployed: reconciled transactions feed the variance base, closed journal entries feed the precedent library, filing packets feed the compliance assistant.
The first deployment handles one bottleneck. The managed model builds from there — one use case per month — until AI covers the full finance function. Not a collection of tools; a coordinated operating layer.
Each deployment leaves a reusable, governed capability that connects to the assets already in place.
We build the AI system on your existing platform and deliver a working solution with documentation and governance in 4–6 weeks. Your team has a live tool and a measurable outcome before the end of the first month.
One new AI system per month, built sequentially across the finance function. Each deployment connects to the ones already live — reconciled transactions feed the close checklist; closed journals feed the reporting narrative. The manual workload drops continuously.
AI covers transactional processing, reconciliation and close, and reporting and strategy. Accountants close the books instead of chasing data; analysts investigate variances instead of building comparison spreadsheets by hand.
No new software subscriptions unless your current platforms genuinely do not support the use case.
Copilot Studio enables agent-building and workflow automation inside your existing environment — typically at no extra platform cost. Excel, Outlook, and SharePoint integrate natively for reconciliation workbooks, invoice inboxes, and close documentation.
Gemini Enterprise integrates AI into Sheets, Gmail, and Drive, with Gemini Studio for custom agents. Reconciliation templates in Sheets, invoice threads in Gmail, and close documentation in Drive connect directly.
For use cases requiring precise document processing at scale — invoice PDFs, bank statements, tax filing packets — we build on Claude. Particularly effective for AP capture, reconciliation matching, and structured reporting from unstructured documents.
For routing between systems, triggering approvals, and connecting bank feeds to the general ledger — sized to your existing stack.
Every system is scoped from the start to comply with the AI regulations of your operating geography. Not a separate workstream, not a post-deployment audit — part of the build.
We map every deployment to its risk classification, document intended use and governance controls, and maintain the records required as obligations phase in through 2026–2027. Outputs feeding statutory reporting get additional controls from design.
Enacted March 2025 under the National AI Agency (ANIA). Our deployments align to its requirements — documented purpose and scope, transparency with end users about AI-assisted outputs, and data handling consistent with national standards.
We align implementations to the NIST AI Risk Management Framework — mapping systems against the four core functions, implementing transparency and accountability controls, and documenting risk decisions for contract and compliance review.
We do not hand over a system and a hope. We hand over a system and its governance documentation.
| Liorant | In-house hire | Large consultancy | Self-managed tools | |
|---|---|---|---|---|
| Working AI in 4–6 weeks | Yes | 3–6 months to hire & onboard | 3–12 months | Immediate, but no governance |
| Finance-specific use cases | 8 pre-built frameworks | Built from scratch | Generic enterprise frameworks | None — you maintain everything |
| Built on your existing platforms | Microsoft, Google, or Claude | Depends who you hire | Depends on the team | Disconnected tools |
| Leadership outcome reporting | Monthly narrative | Depends on config | Separate stream, added cost | None |
| AI compliance (EU / US / SV) | Yes | Needs external counsel | Separate workstream | Full liability on you |
| Ongoing monthly optimisation | Yes | Full-time staff cost | Requires a new engagement | Manual |
| Multilingual delivery | Yes | Depends on hiring | Depends on the team | Not applicable |
The build-vs-hire decision: hiring an internal AI capability takes months and carries the full cost of employment, management, and onboarding. Retaining Liorant delivers a working system in weeks at a fraction of the annual cost — with the flexibility to scale scope as finance priorities change, whether that's a two-person accounting function or a multi-entity finance department.
The activation project delivers a working AI system in 4–6 weeks from signature. Most clients have their first workflow running before the end of week four.
In most cases, no. We build on Microsoft 365, Google Workspace, or Claude — whichever you already licence — and run a platform qualification check before scoping. If a licence change is needed, we tell you at the start so you can decide.
Both. The activation model is the same — we map the workflow, build on your platform, and deliver a system in 4–6 weeks — whether that's a two-person accounting function or a multi-entity finance department processing thousands of transactions a month. What changes is the scope of the first use case, not the model. The one thing this offering doesn't cover is banking, fintech product, or treasury/trading systems — that's a different kind of work, regardless of company size.
Informal use rarely produces consistent, auditable outputs. The difference is implementation: a governed workflow, tested prompts aligned to your document formats, and a system that produces the same quality regardless of who runs it. We do not give you a tool — we give you a process.
With a managed retainer we add one new use case per month, building toward full coverage of the finance function. Each deployment connects to those already live. You also receive a monthly business outcome report and a 90-day opportunity backlog reviewed at each quarterly review.
All implementations use your organisation's existing security perimeter — Microsoft or Google data governance, access controls, and retention policies apply — and we do not move client data outside your environment. Deployments are scoped to GDPR and the EU AI Act, the AI Promotion Law (Decreto 234) and ANIA, or the NIST AI RMF as applicable, with compliance documentation delivered as part of every engagement.
The activation project is a fixed-price engagement with no ongoing obligation. If you continue with a managed retainer, the initial term is three months — after which you can cancel with 30 days' notice.
Yes. Liorant has a presence in Barcelona, Medellín, and San Salvador, with native bilingual delivery in Spanish and English, adapted to each market — including EU AI Act requirements for European clients and market-specific platform availability for Latin American clients.
Your accountants, controllers, and analysts are doing work AI can handle. Liorant builds and runs the systems that give that time back — starting with the bottleneck that costs you the most. A free 30-minute discovery session, no slides, no pitch.
Book your session