We use cookies to make the site work and, with your consent, for analytics and marketing. You can change your choice at any time at the bottom of the page.Cookie statement
NEXERA CoreAI
Businesses increasingly understand what AI can do for them. There is no need to work out how to set it up. In NEXERA Core, agents work on the same data, with the same permissions and the same audit trail as your team. We build it in from the start.
Digital colleagues in NEXERA Core take over the repetitive work: reading invoices, monitoring integrations, checking data and answering the phone. They explain what they find, with advice or an action ready for approval. That opens up a new world: your team has time left for customers and decisions.
Invoices, integrations and data, all day long.
What is out of line and why, in plain language.
What the solution is, and who picks it up.
Prepares the action. A person approves it.
Picks up, recognises the caller and answers questions live from NEXERA Core. Runs on the same voice technology that Starlink uses to speak to its customers worldwide in several languages.
A separate AI tool does not know who you are, what you are allowed to see or what changed today. LIV does.
Ask a question in plain language. LIV finds the answer with the same functions as the dashboards and publishes a report with a summary, insights, charts and tables.
On the right is a replayed session with demo data. The function names are those LIV itself uses.
Choose a question above.
Agents in the core
They read from the same modules as your team, and work within the same rules.
Agents do the groundwork. Your team keeps the final say.
The agents
Hover over an agent for an example
Answers questions about customers, orders, payments and invoices with a report.
From question to report in one step
Which customers are in arrears? LIV retrieves the aged debt analysis, looks up the customers and publishes a report with the largest amounts at the top.
Summarises a contact before a conversation: revenue, payment behaviour and what is outstanding.
Every conversation well prepared
Before a visit, you request the briefing: revenue, payment behaviour and what is outstanding in a few lines, on your phone too.
Reads an answer aloud, for example on the road or on the shop floor.
Answers without reading
You can have an answer on the screen read aloud. That is recorded too, just like the question itself.
An overview of what stands out in the figures, without anyone having to ask.
See sooner what needs attention
Instead of searching yourself, you receive a short list of what departs from the usual pattern, with the figures alongside.
Reads purchase invoices, recognises the supplier and classifies the document, through a specialist partner.
Only the check is left to do
An invoice comes in, is read and prepared as a purchase invoice. A member of staff checks it against the original and posts it.
Revenue, margin, operations and a trend line ahead, from the same definitions as the rest of the system.
Figures that match everywhere else
A trend line is called a trend line, not a forecast. A figure without a source stays empty instead of being estimated.
An agent never gets more permissions than the person asking the question. These are the limits, set in advance.
LIV answers solely on the basis of what the functions return. If it is not there, LIV says so.
The agents do not change orders, postings or master data. At most, they record a note or task, or prepare a proposal. Changes stay with people, through the usual screens and approvals.
An agent sees only what the role of the person asking may see, checked by the same functions as the screens.
The question and the token usage go into the audit trail, as does every instance of something being read aloud.
A question may take no more than eight steps. That keeps costs and response times predictable.
An answer is a report with a fixed structure. A partial answer cannot break the screen.
VAT, postings and payments remain rules that always do the same thing. AI explains and prepares, but does not post.
A figure without a source stays empty. A trend line is called a trend line, not a forecast.
Our DNA
We work with it ourselves every day. That is why you do not need to work out how it is done.
We give our AI agents a clear vision, so that they help build NEXERA Core. That makes us faster, but every change passes the same gates: review, tests and a deliberate step to production by a person. By making the best use of technology, we can offer solutions that were not possible before. All to optimise your business.
AI REVIEW
Every change gets an AI review as a mandatory check, alongside a code scan, human review and more than 10,000 automated tests.
Agents
Agents do the hands-on work, from code to migration, under supervision. Direction and decisions stay with people.
Oversight
A finding about money, security or compliance blocks the merge. Only what a person approves goes to production.
LIV is not tied to a single model. For each customer, we choose the model that suits the work and your data requirements. By default, that is Claude from Anthropic, through a European cloud region, so that your data does not leave the EU. If a customer wants a different model, or a model within their own environment, we set that up. For each question, the model receives only the data that the functions return, and it publishes its answer as a report with a fixed structure.
No. LIV reads through the same functions as the dashboards and does not change orders, postings or master data. On the phone, LIV records only the call and any tasks on the contact card. Changes remain the work of people, through the usual screens and approvals.
LIV sees what the role of the person asking may see. The functions check the same permissions as the screens, and every question goes into the audit trail.
Storage stays in the EU, and questions to LIV are processed in a European cloud region. Which model and which region your environment uses is set out in the data processing agreement. Two things run through the US: signing in, with the identity provider under the EU-US Data Privacy Framework, and speech for AI telephony and reading aloud. Speech is processed in real time, by xAI for AI telephony and by a separate speech service for reading aloud; the transcript and summary are kept in your own environment in the EU.
No. LIV uses the data model, roles and audit trail that are already in place. We switch it on when you are ready.
Because a posting, VAT calculation or payment must always come out the same. There, we use rules. We use AI where it helps: explaining, summarising, flagging and preparing.
NEXERA Core
In a demo, we ask LIV questions about processes like yours, and show how agents work within the permissions and approvals.