How it Works Features The Agents Technology Pricing Meet Your Agent Book a call
Agents for every budget, live in weeks.

Your People are
too valuable for
repetitive work.
Let your Agents take the Load.

We automate all recurring reports, controls, filing preparations and complex reconciliations with custom-built Agents working alongside your existing systems. No ERP replacement. Your team reviews and approves only. Agents are 100% owned by you, run on your IT Stack, and work under your terms. Delivered in weeks, not months.

Credit Control Assistant — liveAV89.ai agent
Credit Control Assistant — AR dashboard live on screen
Execution with no limits.
The Problem

Every high-performing team
has the same problem.

"The work that matters keeps losing to the work that's just urgent."

The Urgency Trap
While fires get fought one by one, the work that truly matters quietly piles up.
Always next, never now.
Up to 55%
of finance teams' time is spent on manual processing.
APQC & ScottMadden benchmarks. Best-in-class reaches 25–35%. Most teams operate somewhere in between.
3+ days
For Board pack preparation, every quarter.
From raw data to final PPT. 12+ days a year on one deliverable.
0 hrs
Real-time visibility on aged receivables.
Until someone manually builds the report, and sends it. By then, it's already out of date.
2–5 days
Lost to reconciliations, every month.
Bank, bordereaux, payroll, SOAs — line by line, every single month.
Weeks
From quote to policy set-up.
Quote accepted. Data emailed. Entered manually. Errors flagged. Corrections requested. Re-entered. Bordereau mapped, rejected, reformatted, resubmitted. Policy live — weeks later.

Remove the burden from your team once and for all. A minimum level of automation is now within everyone's reach — the baseline to have in place before discussing any broader transformation.

Meet your agent

See what your agent actually produces.

A real agent example — the Credit Control Assistant. It takes your raw files from every source, reconciles and controls them, then delivers the finished work: chaser emails ready to send, the full board pack, and an always-ready Credit Committee file with minutes. No more manual updates. Structured, auditable, repeatable — built to your workflow.

The output — what your agent delivers, in one click
The full board pack — all five PPT pages generated by the agent
The full AR Dashboard — 5 pages, one click. Overview, KPIs, ageing waterfall, bucket detail, client exposure, last action plan by ownership & an updated 3-month cash-inflow forecast based on collection patterns and due dates.
Dunning chaser email draft in Outlook
Dunning chaser — drafted in your Outlook. Right tone, right amounts, right policy references. Review and press Send.
Committee minutes drafted in Outlook
Committee minutes — drafted automatically. Minutes with the action tracker, waiting in Drafts.
Live Credit Committee file — decisions, owners and promised dates, carried run to run
Live Credit Committee file — no more manual updates. Decisions, action-item owners, promised dates and comments auto-saved and carried from one run to the next. Minutes and action items sent automatically after each meeting.
How it gets there — with controls you can see
Verify input — full reconciliation and cross-file matching, errors and anomalies flagged
Verify input. Full reconciliation and cross-file matching before anything runs — errors and anomalies flagged upfront, with severity and suggested fix.
Data-integrity controls — 11 of 11 checks passed
Controls — 11 / 11 checks passed. Every figure recomputed from raw data, every output verified — a full audit trail on every run.
Business rules — dunning windows, ageing thresholds and KPI targets, visible and editable
Your business rules — visible and editable. Dunning windows, ageing thresholds, KPI targets. Nothing hidden in code — edit, save, re-run.

Want to see more — and play with the agent yourself? Get into our sandbox, free.

Play with the agent — free →
The Agents

One team of specialists. Pick yours.

Seventeen automation agents, grouped by function. Every one built on the same principle: your team stops producing and starts deciding.

Finance & Operations · 6 agents
Agent 01 · Finance & Operations
Credit Control Assistant
Pain
Premium chasing is manual, inconsistent and person-dependent — overdue receivables pile up while your team is buried in production work.
Gain
Automated dunning sequences, live aged-receivables visibility and escalation logic — recover 15–25% more premium from day one.
Agent 02 · Finance & Operations
SOA & Complex Reconciliation Assistant
Pain
Broker SOAs and bordereaux arrive in every format — days of manual line-matching against internal records each month.
Gain
Any format extracted, references matched, only differences flagged for review — 3–5 days back per month.
Agent 03 · Finance & Operations
Bank & Premium Reconciliation Assistant
Pain
Bank vs GL vs registers matched line by line each close — exceptions surface late, when they are hardest to fix.
Gain
Matches auto-confirmed, only exceptions reach your team, and it keeps learning — 2–4 days back per month.
Agent 04 · Finance & Operations
Management Reporting Assistant
Pain
Month-end pack takes 1–3 days; commentary is a highly iterative process with deep dives across working files, under pressure.
Gain
Board-ready pack in consistent format with full analysis and commentary in 15 minutes — 2–3 days back every month.
Agent 05 · Finance & Operations
Cash Flow Forecasting Assistant
Pain
13-week forecast assembled by hand from multiple sources — slow, error-prone, no real-time treasury view.
Gain
Rolling forecast updated daily from live data, with scenario modelling and variance alerts — live treasury visibility.
Agent 06 · Finance & Operations
KYC Processing Assistant
Pain
Compliance hours lost to data gathering, document verification and report writing — onboarding bottlenecks build up.
Gain
Extraction, rule-based flags and structured output in minutes, with a full auditable decision trail — faster onboarding.
Commercial · 4 agents
Agent 07 · Commercial
Sales Personal Assistant
Pain
Commercial teams still spend a lot of time on admin work — tracking, updating and preparing commercial reviews.
Gain
Automated pipeline tracker with live commercial KPIs; commercial reviews ready in one click — less idle time, more commercial time.
Agent 08 · Commercial
Underwriting Assistant
Pain
Quote information gathered manually across systems — RUW guidance applied inconsistently from case to case.
Gain
Quote information gathered and analysed against your RUW guidance, reports ready — faster, consistent decisions.
Agent 09 · Commercial
Relationship Manager Assistant
Pain
Relationship managers spend too much time gathering information across systems before every renewal.
Gain
Dynamic renewal tracker with portfolio review and performance dashboards — more time for client visits.
Agent 10 · Commercial
Client & Broker Meeting Assistant
Pain
Data from multiple sources (policy, SOA, claims…) means even simple slides take hours to prepare.
Gain
Tailored packs — portfolio performance, claims summary, activity trends — generated before the meeting. Always ready for any client meeting.
Finance & Regulatory · 6 agents
Agent 11 · Finance & Regulatory
Regulatory Reporting Preparer
Pain
CBUAE Forms 1–17, DFSA PIB and Solvency returns need days of manual assembly and cross-checking — teams permanently behind the curve.
Gain
Auto-populated returns with data lineage and full audit trail — filed on time, every time.
Agent 12 · Finance & Regulatory
IFRS 17 Data & Grouping Preparer
Pain
Contract boundary checks and cohort grouping done by hand each close — slow and hard to evidence to auditors.
Gain
Automated GIC formation by cohort with completeness checks and full audit trail — faster close, auditor-ready.
Agent 13 · Finance & Regulatory
IFRS 17 Measurement Preparer
Pain
CSM amortisation and PAA period-end balances live in fragile spreadsheets — one formula away from error.
Gain
Deterministic measurement engine: same inputs, same output, every period — zero formula risk.
Agent 14 · Finance & Regulatory
IFRS 17 Disclosure Preparer
Pain
Rollforwards, disclosure tables and movement notes rebuilt manually every reporting period.
Gain
Auto-populated disclosure tables, rollforwards and movement analysis in Excel or Word — 2–3 days back per quarter.
Agent 15 · Finance & Regulatory
IFRS 9 ECL Preparer
Pain
Staging, PD/LGD/EAD inputs and ECL per exposure maintained in spreadsheets — slow and hard to audit.
Gain
Auditable ECL per exposure with portfolio aggregation and movement schedule — minutes, not days.
Agent 16 · Finance & Regulatory
Financial Statements Preparer
Pain
Financial statements re-typed in Word from the trial balance every cycle — notes chased one by one.
Gain
P&L, balance sheet, equity, cash flow and notes auto-generated from TB and PY FS — 3–5 days back per cycle.
HR · 1 agent
Agent 17 · HR
Payroll & HR Assistant
Pain
Payroll, DEWS, pension and benefits still manual — event handling and multi-step validation before every sign-off.
Gain
100% automated payroll with variance check report and 4-eyes control — your team reviews and approves only. Zero payroll errors.
Work Automation Agents can free up to 8 hours per week for the concerned function permanently.BCG Your team stops producing and starts deciding.
Technology

Is this AI or automation?

Both — but they're structurally different, and the distinction matters for compliance, pricing, and what you can promise your board.

Tier 1
Deterministic Agents — Pure Automation

These agents are not AI in the LLM sense. They are Python scripts and rule-based logic that execute a defined sequence of steps — extract, calculate, format, flag, output — with zero variability. If the input is the same, the output is always the same.

No language model. No probabilistic output. No hallucination risk. Every step is traceable, auditable, and repeatable by design.

What this means in practice: all your Reconciliations, Dashboards, AP and AR routine, Regulatory Returns, Payroll calculations, etc. — all can easily be prepared with deterministic code. And you can audit every number back to its source.
Tier 2
AI-Augmented Agents — Automation + Reasoning

These agents have the same deterministic core as Tier 1 — the data extraction, calculation, and formatting all work the same way. What's added is a reasoning and drafting layer powered by an LLM: written commentary explaining variance drivers, risk narratives, anomaly interpretations.

The LLM does not touch the numbers. It reads the deterministic output and drafts language around it. The figures are always produced by code; the LLM writes the explanation of those figures.

What this means in practice: risk classification rationale, complex input formats standardization, commentaries, etc. — all managed by the model, and reviewed and approved by your team before anything goes out.

The bottom line: Tier 1 is pure automation — code that runs a fixed process. Tier 2 adds an LLM on top, but only for the language, never for the numbers. In both cases, your team reviews and approves the output. Nothing goes to your board or your regulator without a human sign-off.

Under the hood
An open-source, auditable stack. No black boxes.
Deterministic core · every agent
Python 3.12pandasNumPyopenpyxlxlsxwriterpdfplumberTesseract OCRRapidFuzzpython-docxpython-pptxPostgreSQLSQLiteApache AirflowDockerGit
Reasoning layer · Tier 2 only
Llama 3 (self-hosted)MistralQwenvLLMOllamapgvector (RAG)Pydantic validationstructured outputsguardrailshuman sign-off workflow
Every component is open-source and industry-standard. Your IT can inspect every line. Agents are packaged in Docker, versioned in Git, and run entirely on your infrastructure — nothing proprietary locks you in.
Security & compliance
Built for regulated environments.
Data residency by design
Agents run entirely on your infrastructure. Your operational and financial data never leaves your servers.
No vendor access
We never hold, host or access your data. There is nothing on our side to breach.
Full handover at go-live
All code, prompts and BRD documentation are yours from day one. No dependency, no lock-in.
Immutable audit trail
Every action logged with timestamp, data source and applied rule — regulator- and auditor-ready.
DPIA documentation
Complete data protection impact assessment documentation provided with every deployment.
Continuity guaranteed
If we disappeared tomorrow, every agent keeps running. You hold the code and the documentation.
Key immediate benefits

What changes from day one.

01
Capacity created across functions
The recurring workload runs itself — days come back to every team, every month.
02
Focus shifts to value-added work
Your people analyse and decide instead of extracting, matching and formatting.
03
No more repetitive tasks
Reports, reconciliations, filings and chasing run without anyone touching them.
04
Controls strengthen while effort drops
Every run is cross-checked and logged — fewer hands, more assurance.
Real-time data and lean workflows eliminate unnecessary friction and allow better controls and monitoring.
Process

From brief to live Agent in 4 weeks.

A simple, structured process. Your total involvement: ~3 hours.

1
Week 1
The Brief
Tell us where you need to free capacity. Describe your current process. Share your anonymized raw data.
2
Week 2
Build
No involvement needed. We build against your exact data format. Tested against your real scenarios. 100% working with your current system extractions.
3
Week 3
Testing & Final Refine Before Go-Live
One 45-min demo call. You review live output. Give feedback. Nothing goes live until it's right.
4
Week 4
Deploy & Handover
IT runs one command. 30-min handover. All code, prompts & BRD docs handed to you. Performance dashboard live from day one.
Week 1
The Brief

You do

  • Tell us where you need to free capacity
  • Describe your current process
  • Share your anonymized raw data as-is: PDF, XLS, Word, TXT — this is a Standalone Agent (reads files your team already exports — no system connection needed)
  • Or tell us where to fetch the data directly — this is a Connected Agent (integrates with your system via MS Task Scheduler, or system schedules / input folder configuration)

We do

  • Scope the agent from your brief
  • Design workflow & validate data quality issues to handle
  • Send spec for your approval
Your total time commitment: ~3 hours across 4 weeks.
Pricing

Two types of agents.
Two honest price structures.

Deterministic agents: choose between a one-time perpetual fee or an annual license. AI-augmented agents are scoped per engagement. Support billed separately.

Choose your investment structure
Perpetual · Capex
Own it outright
One-time fee within the USD 7,000–20,000 range (AED 25,700–73,500), fixed at scoping. Yours forever, no annual obligation. Typically amortised over 3 years.
Support · 1-year +AED 3,199/mo
Annual License · Opex
Subscribe instead
One-time deploy fee (≈25% of the perpetual price) + annual license (≈40% of the perpetual price per year). Walk away at any renewal.
Support · 1-year +AED 3,199/mo
Support & Updates · Opex
AED 3,199/mo · ≈ USD 870
Output updates · Regulatory change incorporation (DFSA/CBUAE) · Monthly performance & ROI report · File format updates.
With either structure · 1-year engagement
Worked example — a typical AED 40,000 agent
Structure Year 1 Year 2 Year 3 3-Year Total
Perpetual (Capex) AED 40,000 AED 40,000
Annual license (Opex · deploy + license) AED 26,000 AED 16,000 AED 16,000 AED 58,000
Support (either structure · 1-year) +AED 3,199/mo
Annual is 35% less in Year 1 — a smaller approval, classified as Opex. Over 3 years it costs ~45% more: that's the price of flexibility — no capital commitment, walk away at any renewal. Same ratios apply across the price list below.
Block A · Perpetual or annual license
Deterministic Agents — Price Range
Pure Python/RPA · No AI model

Data extraction, calculation, formatting, reconciliation, rule-based logic, report assembly. No language model involved at any point — fully deterministic, fully predictable output every time. Each agent is customized to your systems, data formats and scope — that is why we quote a range, not a price tag. Your exact price is fixed at scoping, before you sign.

Price range
USD 7,000–20,000
AED 25,700–73,500 · fixed at scoping
Get your fixed quote →
100% data sovereign, unconditionally. All deterministic agents run entirely on your own infrastructure. No language model, no external API call, no data leaving your servers. Not a contractual promise — a technical fact.
Block B · Capex — one-time build investment, scoped per engagement
AI-Augmented Agents — Scoped Range
Deterministic core + LLM reasoning layer

Same deterministic core as Tier 1, plus a reasoning and drafting layer powered by an LLM — commentary, narrative, risk synthesis, anomaly interpretation. Priced as a range because reasoning depth, validation scope, and model configuration vary by client and use case. No fixed number until scoped.

What the AI layer actually does — the numbers stay deterministic, the AI does the reading and writing
  • +It writes the commentary on your numbers — board packs, regulatory returns and forecasts arrive with the explanation already drafted: what moved, why, and what deserves your attention. Management Reporting · Regulatory Reporting · Cash Flow Forecasting
  • +It reads documents like a person, at machine speed — quote information gathered from slips and attachments, structured and checked, quote file and report prepared. Quote Preparer Assistant
  • +It makes the judgement call — and documents it — risk factors weighed against your pre-defined RUW guidance into a clear accept / refer / decline recommendation, with a written rationale your underwriter can challenge. Underwriting Assistant
  • +It handles the processes no template fits — reading, interpreting and explaining across many sources — structured or not. If the process exists and requires intelligence, we automate it. Complex bespoke agents
Indicative range
USD 20,000–50,000
AED 73,500–183,500 · scoped per engagement
Build fee only — excludes infrastructure provisioning: model serving (GPU or API), storage and network stay on your stack, sized and controlled by you.
Request a Scope Call →
Data sovereignty — Tier 2 only: The LLM layer runs on infrastructure agreed at scoping — either a self-hosted open-source model on your own servers, or a client-approved external API. This is confirmed per engagement before contract. Tier 1 agents remain 100% data sovereign with no model involved, regardless.

Deterministic agents: USD 7,000–20,000 (AED 25,700–73,500), perpetual or annual license — exact price fixed at scoping · AI-augmented agents scoped per engagement · Support billed separately at AED 3,199/mo, 1-year engagement · Code owned by you from day one on perpetual · Regulatory updates included in support

Why Us

Built by people who know complexity from the inside.
For every team still working through it.

We've worked inside complex organizations and we know most often legacy systems can't be ripped out. We've been there and we build from there. There is always a way.

  • The cheapest, most cost-efficient route to immediate transformation results. No complex or unreliable AI programs, no consulting marathon — working agents, measurable from month one.
  • One-time Capex or annual license. Straight to your problem. Scoped, bounded, standalone. No open-ended consulting engagement.
  • 100% data sovereignty. Your infrastructure. Your control. Deterministic agents run entirely on your own IT stack — your data never leaves your servers. Not a contractual promise, a technical fact. AI-augmented agents default to a self-hosted model; an external API only if you approve it at scoping.
  • Open-source stack. No license fees. Nothing proprietary. Python, PostgreSQL, Docker, Git — industry-standard components your IT can inspect line by line.
  • You own everything. No lock-in. No dependency. All code, prompts and BRD docs handed to you at go-live. If you stop tomorrow, every agent keeps running. AED 0 redundancy cost.
  • Live in 2–4 weeks. Not a 6-month implementation project. Simple agents in 2 weeks, complex deployments in 4. Your total involvement: ~3 hours.
  • Built by insurance CFOs. No briefing required. ACORD, IFRS 17, Solvency, RI bordereaux, CBUAE/DFSA returns — we've been in that room and filed those returns. Adaptable to other regulated industries.
  • GDPR and CBUAE by architecture. Data residency by design, full immutable audit trail, complete DPIA documentation.
  • Low maintenance. High performance. 24/7 monitoring, regulatory updates and accuracy tracking included in support — issues fixed before you notice them.
Yacine Mabrouk, founder of AV89.ai
Who is behind AV89.ai
Yacine Mabrouk · Founder — Insurance CFO

23 years of ground experience in complex and regulated environments across GCC and Europe: AIG, PartnerRe, Marsh McLennan, Coface, Etihad Credit Bureau… I led transition and transformation programs and set up Finance and Operations teams almost from scratch — 23 years of migrations and IFRS (or broader reporting dimensions) updates almost every quarter. Lately, with no-code AI, this is where I delivered my all-time greatest capacity gains on Operations and Finance workflows. A real game changer — zero excuses anymore for inefficiency!

Connect on LinkedIn
Start with an audit

Two ways to begin.

Get started
Free Sandbox

Play with a real working agent yourself, right in your browser — no install, no cost, no commitment.

Play with the agent — free →
For CFO / COO & Board
Deep Audit

A structured diagnosis of where your finance and operations functions leak time, cash and control — with a prioritised roadmap and a board-ready, fixed-price build case. Fee fully creditable toward the build.

Request the audit Download the Audit scope

Your ledgers, SOAs, bordereaux, policy and KYC data never leave your environment. Built by a finance leader, not a generic automation shop.

AV89.ai
Next step

Stop producing.
Start deciding.

Start with one agent. For multi-agent programmes, the first agent is delivered as a scoped proof of concept — at our cost.

Step 01
Discovery Call · 45 min
We identify your bottlenecks and scope the right first agent.
Step 02
Sample Data · 30 min
Share raw data — PDF, XLS, Word, TXT. That's all we need.
Step 03
Live Demo · 45 min
Agent running on your data. You approve. We deploy.

Built and designed by an insurance CFO, for teams who know that complexity and capacity are never an excuse.

Book a call →
LIVE
Credit Control AssistantSOA & Complex Reconciliation AssistantBank & Premium Reconciliation AssistantManagement Reporting AssistantCash Flow Forecasting AssistantKYC Processing AssistantSales Personal AssistantUnderwriting AssistantRelationship Manager AssistantClient & Broker Meeting AssistantRegulatory Reporting PreparerIFRS 17 PreparerIFRS 9 ECL PreparerFinancial Statements PreparerPayroll & HR AssistantCredit Control AssistantSOA & Complex Reconciliation AssistantBank & Premium Reconciliation AssistantManagement Reporting AssistantCash Flow Forecasting AssistantKYC Processing AssistantSales Personal AssistantUnderwriting AssistantRelationship Manager AssistantClient & Broker Meeting AssistantRegulatory Reporting PreparerIFRS 17 PreparerIFRS 9 ECL PreparerFinancial Statements PreparerPayroll & HR Assistant