Coming soon in iconicSecurities
Meet Sari.
AI that
understands structured finance.
Sari reads your terms and conditions, extracts the rules that set up a transaction, and proposes the configuration in iconicSecurities. Your team checks and approves everything Sari suggests. Nothing leaves your infrastructure.

See how
Sari works.
What Sari does
A digital assistant built for securitization
teams.

Reads the deal, not just the data.
A hundred pages of terms and conditions become a structured proposal: transaction parameters, eligibility criteria, portfolio guidelines and triggers, ready to review in iconicSecurities.

Shows its evidenve, every time.
Each value Sari proposes carries the passage it came from. Nothing is invented, nothing is unsourced, and where the contract asks for something the platform cannot configure, Sari says so.

Recommends. Your team decides.
Sari extracts and interprets. A person reviews, edits where needed and approves before anything is exported to the system of record.

Runs where your data lives.
Sari runs in closed infrastructure. Your documents are processed where you choose and do not leave your organization.
AI in banking is too
good to ignore. And too
risky to adopt without
controls.
Data leakage
Documents that leave the institution cannot be pulled back. Most AI tools ask you to trust that they will not.
Hallucination
Confident answers with no source behind them. In a waterfall, one invented number is one wrong payment.
Loss of control
Decisions no one can trace or defend. Regulators and
auditors will ask how a
figure was arrived at.
Sari was built to resolve exactly this. Private, evidence based, always under human control.
Capabilities
Everything your securitization team does by hand. Now with a digital assistant.
Deal setup, straight from the document.
Setting up a transaction means reading a long, formal contract, finding every term that matters, cross-checking them and typing them in by hand. Sari does the reading and the first draft. Your team does the checking.
Upload the contract and Sari extracts the transaction parameters, eligibility criteria, portfolio guidelines and triggers it needs to set up the deal. Results arrive in a few minutes as editable tables with an evidence column pointing to the source passage. Approve the values, export them to iconicSecurities or download them as a spreadsheet.
Finding the right clause is half the job. Sari also maps each extracted condition to the configuration options iconicSecurities supports, fills in the template and explains the mapping. If a condition has no matching option in the platform, Sari flags it instead of forcing a fit.
Answers based on your documents.
Transaction work does not stop at the contract. Investor reports, offering circulars, rating reports and spreadsheets all hold answers somewhere in their pages. Sari lets you ask instead of search.
Upload documents and ask questions in plain language. Sari answers from the text, tables and images in them and shows where the answer came from. Several documents at a time, processed locally.
Sari lifts charts out of documents and rebuilds them as interactive figures. Explore the data behind a graph and ask follow-up questions about it.
Regulation, ahead of time.
Regulatory sources publish constantly, and most of it is not relevant to your programs. Sari watches, filters and summarizes so your team reads only what matters.
Sari checks sources such as ESMA for new publications. When one appears, it reads the article and its attachments, writes a summary and posts it to Slack, Teams or email. Ask follow-up questions about the publication in the same thread. Available as a separate package and requires internet access.
Principles
Five rules Sari follows. No exceptions.
Private. Sari runs inside your closed infrastructure. Your documents and your prompts stay in your organization.
Evidence based. Every extracted value and every recommendation carries a reference to the source document, so you can verify it against the original.
Recommends, never decides. Sari extracts, interprets and suggests. It does not make the decision.
Human in control. A person reviews, edits and approves every change before it reaches the system of record.
Measured before expanded. Sari is built and tested against historical, expert-approved examples. Accuracy is measured on each capability before its scope grows.
Why we built Sari this way
Our clients asked us the same three questions in every conversation about AI. How can it help me, where does my data go, and how do I know the answer is right. Sari is our answer: it acts as a digital assistant that always provides evidence and does so locally, with no data being shared.
András VajdaChief Executive Officer, iconicchain
Applying AI to financial workflows is not about choosing a powerful model. It is understanding the domain, designing the pipeline carefully, preserving source references and evaluating the results continuously.
Aleksandr LoginovHead of AI Solutions, iconicchain
Deployment
Runs where your data lives.
Sari runs in closed infrastructure. Documents are processed inside the environment you choose and are not sent to outside services. Your IT team decides where that environment is.
Your infrastructure
On-premise, private cloud or VPC. Full control of data and access, aligned to your internal policies.
Managed service
Fastest route to a first extraction. We run it, you keep the data boundary.
Your model stack
Where your bank has already approved specific models or controls, Sari can be set up to work within them.
PDF and Word contracts · Documents with text, tables and images · Slack, Teams and email for notifications · iconicSecurities
There’s much more. Dive deeper
on our technical pages.
Discover more for your specific situation
Ready to see Sari on your documents?
Experience Sari now.
"*" indicates required fields
FAQ
Questions we get asked often
No. Sari runs in closed infrastructure. Documents and prompts are processed inside the environment you choose and are not sent to outside services.
Sari does not answer from memory. It indexes your document, retrieves the relevant passages and reasons over those, and every value it proposes is tied to the passage it came from. Where it cannot find evidence, it says so. Each capability is measured against expert-approved historical examples before release.
Transaction parameters such as tranche sizes and the initial portfolio amount, eligibility criteria for individual loans, portfolio guidelines for the pool as a whole, and triggers such as replenishment stop and subordination events. Each is returned as an editable table with an evidence column.
A few minutes for a typical contract of around a hundred pages. Reviewing the result takes longer than producing it, which is the point.
Sari flags it. The row shows no matching configuration and an explanation, so your team sees the gap rather than discovering it later.
No. Sari produces proposals. A person reviews, edits and approves before anything is exported to iconicSecurities, and the platform remains the system of record for all calculation and transaction logic.
No. Sari is an optional module. Everything in iconicSecurities works without it.
Contracts in PDF or Word for extraction. For questions and chart exploration, documents with text, tables and images in the supported formats.