International applied AI lab

AI for the work a business can’t afford to get wrong.

Ainoova builds AI products for finance teams, for the meetings where decisions get made, and for Europe’s local businesses. All four run on one reliability layer: every output traced to its source, every consequential step approved by a person, every connection read-only unless you say otherwise.

Ainoova reliability layer
01Traced to source
02Approved by a person
03Read-only by default
04Computed, not generated
05Honest when it can’t
06Proven before it ships
Frontier models·your ledgers, inboxes, calls and storefronts
The products

Four products. One standard.

Each one does a job a business already pays people to do, and each one is held to the same rules before it touches real money, real meetings or a real customer.

FinanceIn production

Foresail

FP&A, done for you.

An AI-run FP&A service for finance teams in multi-entity groups. Foresail sets up the model, forecasts and reports every month. Your team reviews and approves.

  • Actuals against budget, balance sheet and intercompany, across every entity and currency
  • Any figure drills down to the accounts behind it, month by month
  • Read-only to your ERP, starting with Xero
For CFOs and finance teamsforesail.app
Finance operationsBy invitation

Accrue

The finance team’s work, kept current by AI.

Every workstream, from a bank onboarding to an audit, a tax question or a forecast, becomes a stream: where it stands, the next step and who owns it. Accrue reads the team’s email, chat and meeting notes and keeps it current, so nobody updates it by hand.

  • Applies a change on its own only when it is sure; otherwise one click in Review
  • Every change shows its evidence and can be undone
  • Read-only connectors: it never sends, replies or posts
For finance leaders and their teamsRequest access
MeetingsEarly access · free

Companion

A real-time meeting coach for Windows.

Companion listens from your own machine, so no bot joins the call, and points at the best next move while the conversation is still live. Afterwards it writes the debrief and the follow-ups.

  • Talk share, silence and monologues measured on your PC, with no model calls
  • Follows the room across languages, even mid-sentence
  • It drafts, you send. Recordings stay on your PC
For anyone whose job happens in meetingsMeet Companion
Local businessPilot · Spain

Sites

Modern websites for Europe’s local businesses.

Plenty of Europe’s best-reviewed local businesses still send customers to a website that is dead, out of date or missing. Sites finds them, audits what they have and builds a fast, mobile-first replacement the owner can see before they decide anything.

  • Audits domain, security certificate, mobile layout and freshness before proposing
  • Proposals are marked as proposals and kept out of search engines
  • Follows each country’s outreach rules: in Spain, a call first, never a cold email
For restaurants, shops and local servicesJoin the pilot
The reliability layer

The model is a component. Reliability is the product.

A business can’t run on an answer that is right most of the time. So the lab owns six rules that every Ainoova product inherits, and each one shows up in what we ship, not in a policy page.

Rule
Foresail
Accrue
Companion
Sites
01

Traced to source

Every output points at the evidence behind it, so anyone can check it in one click.

Any figure opens into its accounts, then each account by month.

Every stream update links back to the email, chat or meeting that caused it.

A contradiction flag always names the earlier line it disagrees with.

02

Approved by a person

AI does the work. A person signs off whatever has consequences.

Your team reviews and approves the forecast and the report each month.

Uncertain changes wait in Review, and a person’s own edit is never overwritten for seven days.

Nothing goes live without the owner’s yes.

03

Read-only by default

The least access that does the job. Writing back is the exception, never the default.

Read-only to the ERP. Xero tokens are encrypted and never reach the browser.

Connectors only read. It never sends, replies or posts.

Lookups run through a read-only allowlist that fails closed. Sending takes two deliberate clicks.

04

Computed, not generated

Language models handle language. Arithmetic, rates and thresholds stay in code.

Currencies convert at ECB reference rates: monthly average for the P&L, month-end close for the balance sheet.

Automated health checks look for ledger problems before anyone reads a report.

Talk share and monologue length are measured locally from the audio, with no model involved.

05

Honest when it can’t

When a source or a model is missing, the product keeps working and says what it could not reach.

Without a model it keeps filing work on built-in rules.

Without mail and chat access, briefings fall back to calendar and CRM and list what they could not see.

06

Proven before it ships

New versions meet synthetic data before they meet yours, and roll back fast when something breaks.

Every page, drill-down and export is tested against a fictional group before release, and production is checked after every deploy.

Sign-in requires two-factor authentication. Each company sees only its own data.

The lab

International by default. Small on purpose.

Ainoova is a small, senior team spread across four cities, so the products are covered in the hours businesses actually work. Nobody here manages a team that writes code. Everybody here writes it.

We stay small so that whoever writes a rule is also the one accountable for it working on a live ledger, in a live call, or on a live storefront.

i

Customer zero, first

Foresail and Accrue ran inside a real finance team, on real ledgers and a real inbox, before we offered them to anyone else. We find the failure modes so you don’t.

ii

Built for how Europe works

Several entities, several currencies, several languages, and rules that differ by country. Multi-currency consolidation, multilingual meetings and country-specific outreach rules are designed in from the start.

iii

One lab, shared lessons

A safeguard learned in one product becomes the default in the others. That is the point of building four products on one layer.

San Francisco
 
New York
 
London
 
Barcelona
 
Talk to the lab

Put AI where it has to be right.

Questions about a product, a partnership or a pilot? Write to us and a person from the team will answer.