ibl.ai
ibl.ai sells an AI agent platform that institutions run themselves, in their own cloud, on premise or fully air gapped, with the full code licensed perpetually to the customer. For banks, broker dealers, asset managers and wealth firms it offers agents for compliance monitoring, risk analysis, KYC and anti money laundering checks, fraud detection, client advisory, portfolio analysis, onboarding and trading operations.
The platform works with Claude, GPT, Gemini, Copilot, Amazon Bedrock, NVIDIA and open source models, chosen and switched by the customer. It connects to core systems such as FIS, Fiserv, SS&C, Jack Henry and Broadridge, CRMs such as Salesforce Financial Cloud, Redtail and Wealthbox, and data from Bloomberg, Refinitiv and FactSet. It stresses documented models, version control, independent validation and audit trails for model risk reviews, and says it is SOC 2 Type II compliant.
Enterprise pilots start at $15,000. ibl.ai serves more than 400 organizations across education, government, healthcare and finance, though it names no financial institution. It is family owned and based in New York.
Capability Axes
Capability grades
15 of 15 axes rated · 11 graded A or B
Model serving, orchestration and ready built agents make up the platform, and the financial services agents for compliance, risk and financial crime are shipped products rather than plans.
Agents reach into trading operations and financial crime work, yet no approval threshold, escalation path or limit on what an agent may act on is described.
Models are documented, version controlled and independently validated, and reasoning can be reproduced from audit trails a bank can export for examiners.
More than 400 organizations and 1.6 million users sound substantial, but the names attached, such as Syracuse University and Google, include no bank, broker dealer or asset manager.
Run air gapped with no internet route, the platform gives the vendor no path to customer data, and customers keep the code under a perpetual license.
Zero third party access to client data is stated, which keeps the data with the bank.
SOC 2 Type II compliance is stated, without a report or trust center to back it.
Federal Reserve model risk guidance, SEC and FINRA supervision and SOX are all addressed in the financial services material. These are alignment claims for a software supplier, not an admission to any regime.
Financial crime screening is among the prebuilt agents, and screening is known for uneven impact across groups. Fairness testing isn't described for any of them.
The institution owns the deployment and every decision it makes, and ibl.ai publishes no guarantee or indemnity if its code or agents cause an error.
Customers pick and switch among named model families such as Claude, GPT and Gemini, or run open source models with no outside provider at all.
The named integrations reach core banking and data providers, FIS, Fiserv and Bloomberg among them, and a guide covers the Model Context Protocol connection. None is backed by a partner listing.
Deployment can sit in the bank's own cloud or on premise, or run in a managed cloud instead.
A fixed scope pilot starts at $15,000 and integration and deployment run $25,000 to $80,000, so a bank can budget the project. What a self serve credit costs isn't published.
Banks, broker dealers and wealth firms are the financial targets, but they share the platform with education, government and healthcare customers.
Product updates
Material product, regulatory, evidence and commercial changes at ibl.ai, each verified against a live source and tagged to the capability axis it bears on. Funding rounds and awards are not product changes and are not logged.
ibl.ai gave every agent a privacy policy setting, shipped in v2.12.0 on 12 October: an admin can let users decide, make every conversation with that agent incognito, or never allow incognito, alongside the existing PII filter. One API endpoint resolves the agent policy, the organization setting, the chat toggle and the user default, and returns the resulting mode, which tier decided it and whether it is locked.
ibl.ai extended its PII and PHI filtering to in chat file uploads, covering text, Office documents, images (OCR plus pixel redaction) and PDFs (rasterized, redacted and rebuilt) on every path where a file reaches an LLM, with per agent block, redact or allow settings. Privacy mode now also keeps conversation text out of run tracking, the moderation log, application logs, Sentry events and Langfuse traces. A new read only privacy flags API records every PII or PHI detection across uploads, prompts, responses and memory retrieval, storing only the entity type, never the raw value.
Admins can now define named, platform scoped network policies for agent sandbox VMs: a host and port allowlist sent to the sandbox each session, plus a custom egress profile per agent, and a missing or cross tenant policy fails closed. Sandbox secrets can also be drawn from stored integration credentials at runtime so rotations propagate, and sandbox runtime is metered at a configurable default of $1 per 10 minutes.
Compared With
Most editorial comparisons pair two vendors the index assesses as direct competitors for the same buyer. Some pair vendors that are adjacent rather than rival, where the useful question is where one ends and the other begins. Each carries a verdict, the buyer conditions that favor each vendor, and a graded side by side.
Alternatives to ibl.ai
Vendors in the same categories as ibl.ai whose profiles look most alike, each with a line on what the product does. Similar is not the same: check that each one does the job you need. No vendor pays for placement.
AlphaBitCore sells an AI control plane that lets regulated firms govern their AI models, agents, tools and workflows in production.
Xapien automates the research step of due diligence and returns a document rather than a score.
Sedric turns a regulated firm's policies and the rules it operates under into enforceable system logic, then applies preventive, detective and corrective controls across every customer touchpoint.
Block Convey builds two AI governance layers for banks, asset managers, insurers and fintechs.
OnFinance AI runs ComplianceOS, a platform of more than seventy agents that monitor regulator portals, interpret circulars at clause level, assign owners, track deadlines, score controls…
EVE AI Core is a deterministic control plane that intercepts every AI action before execution rather than reviewing it afterwards, returning an allow, block or modify verdict against…
The index publishes no overall score or ranking. See what we assess and the methodology.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | Data Protection Terms | Implementation | Source |
|---|---|---|---|---|
|
Self serve credits free to start. Enterprise pilot from $15,000; integration and deployment $25,000 to $80,000; codebase transfer quoted.
|
Two routes. Self serve use runs on prepaid, rechargeable credits with a ceiling the customer sets and no per seat fees; no credit rate is published. Enterprise engagements are one time fees: a fixed scope pilot from $15,000 with one or two production agents, one integration, deployment on the customer's infrastructure or ibl.ai cloud and weekly engineering sessions, credited toward a full engagement; integration and deployment from $25,000 to $80,000 for multi agent deployment, data integration, single sign on and custom agents; and a quoted codebase transfer with full source code, a perpetual license and a dedicated engineering team. After deployment the customer pays only its own model token and infrastructure costs. | Not tiered on the pricing page. The homepage states SOC 2 Type II, and self hosted, customer cloud, on premise and air gapped deployments are offered. | Enterprise work is priced as one time engagements, and the pilot fee is credited toward a full engagement. | Vendor Published |
Prices taken from the vendor's own pricing page as of 3 October 2026, in US dollars. The credit rate on the self serve route is not published, and neither are the expected model and infrastructure costs after deployment.