People and companies
Add the individuals and companies named in a document and link them to the right structure.
Vericia AI
Vericia AI works inside your records. It answers questions with its sources shown, reads client documents, drafts minutes and risk assessments, and prepares the record changes that follow. Nothing changes until one of your people approves it.
Mr Tan resigned as a director yesterday. What needs to happen?
Ask
Open the assistant beside any page and ask in plain language. It answers from your firm's records and the rules pack for the jurisdiction, links every record it mentions and lists exactly what it read.
What does the Demo group have due before the end of the year?
Read documents
Upload a passport, an identity card, a proof of address or a company document. Vericia AI reads it, compares it with the record and proposes the updates. You tick the changes you accept.
Check before accepting: the name on the document differs from the record.
Prepare changes
Ask Vericia AI to act and it sets out a numbered plan: the records to update, the filings to raise, the requests to send to the client. You review each step, untick what you do not want, and approve.
“Record the board meeting from these notes.” Attach the notes, a signed form or a full onboarding pack.
Each step is listed with the exact change it will make. Nothing has been saved at this point.
Approved steps run under your name and permissions, with the same validation and audit trail as if you had made them yourself.
Add the individuals and companies named in a document and link them to the right structure.
Appoint or cease officers, record holdings and share transfers, with the notices that follow.
Create the meeting, set the agenda, draft the minutes and add the resolutions passed.
Start a draft risk assessment, request a KYC refresh from the client, or record a screening review.
Assign tasks to your team and send requests to clients through the portal.
Open a new draft of a constitution when a resolution amends it.
Everyday work
Nine jobs Vericia AI takes on today. Each one ends the same way: with a draft or a proposal for one of your people to review.
Attach the onboarding pack. Vericia AI identifies the people and companies in it, links them into the structure and lists what is still missing.
It finds expired and outdated documents across a file and prepares the request to the client through the portal.
It compares a sanctions or adverse media alert with the record and drafts the decision note. The decision stays with your compliance team.
A first draft built from the record, with the reasons for each rating, left as a draft for compliance to review and sign off.
From your notes to draft minutes and resolutions, together with the record changes and filings they give rise to.
Resignation letters, transfer forms and amended constitutions are read back into the records they affect.
The rule, the deadline and the form for any filing, with a draft cover letter to go with it.
A short note on your week: what to deal with first and what falls due next, across every entity you look after.
Likely duplicates, incomplete addresses and missing tax details are found across your records, with the corrections proposed.
Your AI apps
Your people can connect the AI app they already use to Vericia through its MCP server. They ask about the firm's records where they work, with the same permissions and safeguards as inside Vericia.
Connected through the Model Context Protocol (MCP), the open standard for linking AI apps to business systems. Switched on for your firm on request.
Safeguards
Regulated firms need to know exactly what an AI system can see and do. Vericia AI operates within the same permissions, masking and audit trail as your people.
Vericia AI never alters a record on its own. Each proposed change waits for the person who asked to approve it.
Every record is read with the permissions of the person asking: the same roles, the same jurisdictions, nothing more.
Tax, identity and bank account numbers are reduced to their last four characters before anything reaches the AI model.
Vericia AI runs on Claude from Anthropic. Anthropic does not train its models on the data Vericia sends.
Each question, each record read and each plan proposed, approved or declined is logged against the person who asked.
Vericia AI never signs minutes and never decides a screening alert. Sign-offs and four-eyes approvals remain human decisions.
Cost control
Vericia AI is optional. It stays off until your firm asks for it, and you decide who uses it and how much.
Enabled firm by firm, and only for the roles your administrators choose.
Usage is capped each month, so spending never exceeds the limit you have set.
Each team receives a share of the allowance, equal or in the proportions you choose. A team that uses its share waits for next month while the others carry on.
Administrators see this month's usage against the allowance, by team and by person, with six months of history.
We agree with you which Claude model your firm uses, balancing depth of reasoning against cost.
Billed monthly on actual usage. If nobody uses it, it costs nothing.
Questions
No. It can read records and propose changes, but every change waits for the approval of the person who asked. Approved changes are made under that person's name and permissions and appear in the audit trail like any other edit.
No. Vericia AI uses Claude through Anthropic's commercial service, which does not train models on the data it is sent. Tax, identity and bank account numbers are masked before they leave Vericia.
A document is sent to the model for that one request and is not retained by Vericia. We record its name, size and digital fingerprint so that it can be matched to the copy in your own document management system.
No. Every record is fetched with the permissions of the person asking. An administrator for one jurisdiction cannot use the assistant to read files in another.
Answers are drawn from your records and the rules pack, not from the model's general knowledge, and each one lists the records it read so that it can be verified. As with any assistant, your people remain responsible for the decisions they take.
Yes. Claude, ChatGPT and Microsoft Copilot can connect to Vericia through its MCP server. They read with the person's own permissions, and any change they propose waits for that person's approval in Vericia.
No. Vericia AI is optional and is switched off until your firm requests it. Every other part of the platform works in full without it.
See it for yourself
Book a 45-minute demonstration. Tell us which jurisdictions you administer, and we will show you Vericia at work on structures like your own.