Rules engines worked when regulation was bounded and binary. Cross-border trade, fragmented jurisdictions, and judgment-heavy regimes have moved the goalposts. statisTicker is the agentic-AI layer for the part of compliance rules engines can’t reach, and stays out of the part where they still can. We start with fiduciary and trust companies, where IPS-driven mandates contain the judgment-heavy exceptions a rules engine cannot safely resolve on its own.
Contact the founding team →The first generation of compliance technology was built around a stable picture: one institution, one regulator, one set of rules updated on a predictable cadence. A restricted-list check returned yes or no; an OFAC screen ran against a single sanctions register; an employee-trading clearance came back in seconds. That world is still alive in equity pre-trade. But across the rest of compliance, the rule volume per transaction has grown faster than any institution has scaled reviewers to read it.
Video could not be loaded in this browser.
The gap between the two lines is the point.
Two properties decide whether agentic AI earns its keep in a compliance workflow. The first is rule determinism: how mechanically a regulation maps a fact to a decision. The second is judgment density: how much human interpretation a given case actually requires. Where rules are deterministic and judgment is shallow, rules engines have run the table for decades and there is no AI value to add. Where rules fragment across jurisdictions and judgment is interpretive, the rules-engine ceiling breaks. That is where statisTicker concentrates.
The upper-left quadrant is where agentic compliance AI can add value. Everything else is already someone else’s solved problem.
AI capital in compliance went to catching abuse after the tape, and to reading email. Pre-trade stayed inside the OMS as boolean rules. A trust IPS is neither: “avoid controversial ESG exposure,” “multi-generational risk parity,” multi-beneficiary and multi-jurisdiction constraints. That language does not compile into a restricted-list check. The missing product is a second reviewer that reads the mandate, flags a statistically anomalous order, and holds it before execution — on infrastructure the institution controls, with the CCO setting the threshold.
| Competitor | Their Angle | statisTicker Differentiation |
|---|---|---|
|
Charles River / AIM / Aladdin / Eze
OMS rules engines
|
Pre-trade compliance as 1,700+ hard-coded rules inside the order-management system | They own binary, deterministic checks. They cannot interpret the subjective language of a trust or family-office IPS. The OMS also encodes firm and investment-committee strategies as the same coded rules, alongside client guidelines, and its what-if sandbox lets the committee run those rules and risk models against live portfolios and market data. That is still numbers against a rule library, not mandate prose against a judgment panel. |
|
RedBlack / Blaze Portfolio
RIA IPS / rebalancing tools
|
Client-level IPS encoded as rules during rebalancing and trade generation | Closest overlap on mandates: household, account, and firm restrictions, security equivalents, model weights, and drift bands are checked before a rebalance hits the blotter. Blaze adds SRI/ESG filters of the same kind. The advisor still has to translate IPS prose into those coded overlays. Cloud SaaS, no on-prem judgment panel, and silent on the order that breaks no rule but looks like nothing this book has done. |
|
Nasdaq SMARTS / NICE Actimize / SteelEye
Post-trade surveillance
|
Market-abuse, MAR/MiFID, and communications surveillance after the trade | A different problem. SMARTS and Actimize detect spoofing, layering, and other MAR/MiFID abuse in real time or T+1; SteelEye reconstructs the trade and the chat after the fact. That is misconduct surveillance, not a fiduciary hold. They do not sit in the order path to stop an IPS breach before execution. Sanctions and OFAC screens are a separate, already-solved rules-engine job — not what this stack replaces, and not the layer we occupy. |
|
Behavox / Shield
AI communications platforms
|
Domain models on email, chat, and voice for conduct surveillance | They score email, chat, and voice — 150+ channels in Behavox Quantum, risk scenarios in Shield — for conduct, MNPI, and off-channel communications. Where Behavox also runs trade surveillance, it is market-manipulation detection, same family as SMARTS, not mandate interpretation. They can reconstruct the conversation around a trade. They never read the IPS, and they never hold the order. |
|
Addepar / SEI
Wealth data & trust platforms
|
Portfolio aggregation, trust accounting, post-trade reporting | Addepar is the UHNW/SFO book: entities, ownership, alternatives, and performance across custodians. SEI runs the trust ledger. Addepar Trading, where licensed, adds coded limit checks and rebalancing on that book — the same class of rules as RedBlack, not a second reviewer of an ambiguous mandate. The core platform is a view of the book. It does not learn what this desk normally does, and it does not produce a CCO-tunable hold with an evidence-weighted audit record. |
Adjacent, not substitutes: employee-trading pre-clearance (ComplySci, StarCompliance) covers personal accounts, not client portfolios. AML/KYC (Fenergo, ComplyAdvantage) is client onboarding, not mandate adherence. FundApps, now with SteelEye, is shareholding-disclosure and coded mandate rules — still not a judgment layer on an ambiguous IPS.
We start with fiduciary and trust companies because the product, the IPS workflow, and our regulatory domain expertise meet in one buyer and one problem. Energy trading and mortgage compliance are growth-roadmap markets: the same judgment layer can travel there, but they follow the first wedge rather than compete with it for day-one focus.
IPS-driven mandate compliance involves concentration edge cases, prudent-investor interpretation, ESG carve-outs, and other calls that are not simple rule lookups. This is the worked example for the first deployment.
A single commodity trade can touch multiple regulatory regimes simultaneously. The same judgment layer can extend to CFTC, FERC, REMIT, and desk-specific mandate checks.
Loan files combine high volume with interpretive requirements across TILA, RESPA, HMDA, ECOA, fair-lending, and state overlays. The product shape travels here after the fiduciary/trust workflow proves the deployment and review model.
Growth roadmap: energy trading/ETRM first, followed by mortgage and lending. These are real adjacencies, but out of the initial fiduciary/trust wedge. Out of scope entirely: US equity pre-trade for hedge funds and prop trading (solved by rules engines), commercial REITs, sovereign treasury operations.
Independent models read the same mandate and the same order, on infrastructure the institution controls. Client order flow and IPS text never enter a commercial API. The CCO owns the hold threshold. Every clear or hold ships with an audit-grade rationale an examiner can replay. How the panel is composed, how dissent is resolved, and how the threshold is computed are not on this page. A working session under NDA is the next step.
Open-weight models, run inside the client boundary. No cloud inference on positions, notional, or mandate text.
The institution sets the operating point. We do not bake a vendor opinion into the hold policy.
A signed rationale on every decision, with dissenting views retained. Built to answer “what are you doing on the front end?”
Binary checks stay where they already work. We take the judgment-heavy residue the OMS cannot compile.
The question that comes up in every conversation is token cost. The rest is custody, training, and what the harness is not.
The recurring question
If we run open models on your own hardware: nothing per token. The cost is compute and electricity you already own. If you prefer subscriptions you already hold — Microsoft/Azure, OpenAI, Anthropic — we wire the harness into those under your agreements, with usage estimated per use case before anything runs. Either way, you take on no new metered dependency on us.
A disciplined way of working, not a software platform installed from the cloud. It orchestrates several independent frontier AI models together with deterministic checks, entirely inside your environment. Facts and arithmetic are computed deterministically first; the models never do math. Only genuinely ambiguous judgment calls reach the model panel, and every answer comes back with a stated confidence and a complete audit record of the evidence behind it.
No. Everything runs in your environment, on your hardware or inside your existing cloud tenant. There is no feed to us, no break in chain of custody, and no data-transfer authorization to obtain, because no data is transferred. The internal approvals that make a “free” pilot expensive never come into play.
No. Nothing is trained or fine-tuned on your data, and we retain nothing. The harness is engineering and statistics around models, not a model built from client data.
Keep them. Bright-line rules should stay deterministic, and our first tier blocks those breaches inline the same way. The difference is what a rules engine cannot do. It cannot read prose: the mandate and policy language that today requires a person. It cannot show why an order that cleared was safe to clear; we return a probability of violation with a stated confidence range and an evidence-weighted audit record, which turns “document and let it go” into a defensible file. And it cannot notice the order that breaks no rule but looks like nothing this desk has ever done, because it has no memory of what the desk normally does. Our statistical layer learns that from your own history.
Yes, and it may be the natural first step. The same discipline applies to document-heavy review: deterministic extraction and arithmetic first, judgment cross-checked across independent evaluators, and output that cites its evidence and states its confidence. This is not pasting documents into a chatbot; every conclusion shows where it came from and how sure the system is. We would scope one workflow of your choosing and deliver it as a fixed engagement.
No. The harness runs inside your tenant and can use the Copilot and Azure capacity you already pay for as its model layer. We are not replacing your stack; we are making what you already own do more.
Small and local. You pick one workflow: a KYC file review, an underwriting memo, or a pre-trade check on one desk. We deliver a working result on your infrastructure in weeks, at a fixed fee, judged on two things your team measures: hours returned, and the quality of the audit record. Because no data moves, there is no integration project and no transfer approval on the critical path.
Regulatory taxonomy, a multi-model on-prem harness, and commercial partnerships. The CCO owns the threshold. We provide the engine and the evidence chain.
Persia Shokoohi, J.D.
Compliance & Regulatory
Mash Zahid
Architecture & Build
JD Henao
Growth & Partnerships
Regulatory liability cannot be outsourced. The institution’s CCO owns the threshold, the posture, and the block/clear policy. statisTicker provides the engine, the rationale capture, and the audit-grade evidence chain: the defensible front-end answer when a regulator asks “what are you doing on the front end?”
Persia Shokoohi is a senior compliance, risk management, and governance executive with more than a decade leading regulatory, audit, information security, and enterprise risk programs across investment advisers, trust companies, broker-dealers, and financial institutions. She has served as Chief Compliance Officer for SEC-registered investment advisers, with oversight spanning SEC, FINRA, OCC, CFTC, NCUA, FDIC, and state requirements.
As founder of Compliantry Consulting, she advises domestic and international firms on regulatory compliance, governance, risk management, data governance, regulatory reporting, and market expansion. A graduate of the University of Miami School of Law and Emory University, she was named one of Corporate Counsel Business Journal’s ‘50 Women to Watch’ in 2024 and speaks regularly on compliance, financial data governance, fraud prevention, and regulatory technology.
Mash Zahid builds and deploys AI systems at enterprise scale across financial services, energy and utilities, telecommunications, healthcare, automotive, and professional services. Most recently he led the agentic AI transformation of a global automaker’s operations and HR functions, reporting to its Chief Operating Officer, and has field-deployed multi-model agentic AI in critical power-delivery infrastructure and regulated financial compliance.
As an Associate Partner at IBM, he led large-scale generative AI engagements for Fortune 100 clients, including a conversation intelligence program for the leading telco that returned more than $20M in its first month. As Global Director of AI Strategy at KPMG, he advanced the firm’s flagship audit platform into neural network–based deep learning. He served on the Architecture Review Board for Moveworks before its $2.85B acquisition by ServiceNow. Earlier, his workforce planning work at The Home Depot was the cover story in Harvard Business Review.
He holds an MBA from the University of Chicago Booth School of Business, with doctoral coursework in behavioral finance. In 2025 he delivered a keynote to 6,846 practitioners at OpenAI Academy on scaling enterprise AI through systems thinking, and placed first at a Google hackathon opened by Sergey Brin.
As Head of Growth at statisTicker, Juan Diego Henao leads market development, strategic partnerships, and customer engagement. With a background spanning Fortune 1000 companies, emerging startups, and financial services, he works closely with clients and partners to align the product with real-world operating constraints. His focus is building relationships, creating commercial opportunities, and ensuring engagements are scoped to deliver operational and financial impact.
MBA in Marketing and Entrepreneurship from the University of Chicago Booth School of Business. MA in Organizational Behavior and BA in Economics from Stanford University.