Specialized agents
Purpose-built agents for portfolio work, trading operations, reconciliation, corporate actions, research, and document retrieval. Not one general model asked to do everything.
SPECTER OS BY VIDENT
The AI-native operating layer for ETF trading and asset management
SPECTER OS is our AI-native operating layer for ETF operations. It sits above the systems you already run — connecting portfolio management, trading, risk, operations, research, and market data through specialized AI agents and deterministic financial workflows, so your team stops being the integration layer.
In active development. Built from twenty years inside ETF portfolio management, trading and risk.
THE PROBLEM
ETF operations run on glue
Holdings arrive in one system. Trades live in another. Corporate actions land in an inbox. NAV validation happens in a spreadsheet that one person understands. Baskets, creations, and redemptions move between four systems and two spreadsheets before anyone can see the day clearly.
The work is not hard because the people are bad at it. The work is hard because it is spread across a dozen systems that were never designed to talk to each other, and the coordination lives in your team's heads.
Every firm has built the same shadow infrastructure: the morning checks, the reconciliation, the exception chasing, the audit trail reconstructed after the fact. It is expensive, it does not scale, and it is invisible on any org chart.
WHAT VIDENT IS
A coordination layer, not a replacement
Vident is not another system of record and it is not a chatbot bolted onto one. It is an intelligent coordination layer that sits above the platforms you already run.
Nothing gets ripped out. Vident reads from your existing portfolio management, order management, custody, and market data systems, and coordinates the work across them. Where a calculation has a correct answer, it is computed deterministically, not generated. Where judgement or approval is required, a human stays in the loop, and every step is logged.
The result is that cross-system workflows that currently take a morning and three people take a workflow and one approval.
CAPABILITIES
What Vident handles
Thirteen workflows that are on the roadmap, not thirteen features of a generic assistant.
Portfolio & trading
- ETF portfolio management
- Trading and rebalance workflows
- ETF basket and creation/redemption workflows
Reconciliation & corporate actions
- Holdings and position reconciliation
- Corporate actions and dividends
- Exception detection and escalation
Daily operations
- Morning operational checks
- End-of-day NAV and shadow NAV validation
- Cross-system workflow automation
Risk, research & knowledge
- Portfolio and regulatory risk monitoring
- Research and market intelligence
- Document and institutional knowledge retrieval
- Human approvals and audit trails
HOW IT WORKS
Deterministic where it matters, agentic where it helps
Most AI in financial services fails for one reason: it guesses at numbers. Vident does not guess.
Deterministic calculations
Anything with a correct answer, NAV, exposure, drift, basket composition, is computed, not generated. A language model never invents a number that lands in a report.
Orchestration
Multi-step workflows run across systems and teams with persistent state and memory, so a process that started this morning still knows where it was this afternoon.
Validation agents
Every material output is checked by a second, independent pass before it reaches a human.
Human in the loop
Approvals are configured, not implied. Nothing touches a book without the right person signing off.
Audit by default
Permissions, audit logging, and evaluation systems are part of the architecture, not a compliance feature bolted on at the end.
BUILT FOR CONTROL
Built for the way regulated firms actually work
Financial infrastructure earns trust through its controls, not its demos. Vident ships with role-based permissions, configurable approval chains, complete audit logging of every agent action and human decision, and evaluation systems that measure whether the agents are still right after the model changes.
If a regulator asks what happened, the answer is in the log, not in someone's memory of the morning.
Role-based permissions
Access is defined by role, not by who happened to be copied on the email.
Configurable approval chains
Approvals are explicit and configured per workflow, so nothing touches a book without sign-off.
Complete audit logging
Every agent action and every human decision is recorded and attributable.
Evaluation systems
Continuous measurement of whether the agents are still right after a model or data change.
INFRASTRUCTURE
Cloud or on-premises
SPECTER OS runs as a managed cloud deployment on AWS — Amazon Bedrock for model hosting, Step Functions for workflow orchestration, RDS for transactional state, S3 for documents and position files, Lambda for compute, and CloudWatch for logging and observability.
- Amazon Bedrock
- AWS Step Functions
- Amazon RDS
- Amazon S3
- AWS Lambda
- Amazon CloudWatch
PRIVATE DEPLOYMENT
Runs inside your own environment
Most asset managers cannot send custody positions, holdings and counterparty exposure to a third-party model provider. For those firms, SPECTER OS runs entirely on their own premises — the same agent architecture, against models served locally on GPU hardware, with no data leaving the building.
- NVIDIA Jetson AGX Orin
- CUDA-accelerated inference
- NVIDIA Nemotron
- On-premises
- No data leaves the firm
SERVICES
Vident also works directly with asset managers
Alongside the platform, our consulting and technology teams work hands-on with asset managers on AI strategy, agentic workflow design, data platform modernization, and governance. It is how we learned these workflows in the first place, and it is often how a relationship with Vident starts.
Building for ETF operations?
SPECTER OS is being built from twenty years inside ETF portfolio management, trading and risk. If you run an ETF operation and want a say in how this develops, we would like to talk.
Start a conversation