CipherOwl

The removal test leaves public data anyone can download for nothing. Blockchain ledgers are open by construction, so the raw material carries no value on its own and the entire product is the interpretation: machine learning driving transaction monitoring and predictive risk assessment, agents reasoning over ledger activity, and automated production of investigation and reporting output.

The company describes itself as an intelligence layer rather than a data provider, and the distinction is accurate, because what it sells is the reading of information its customers could already obtain.

The oversight commitment is stated as a design property rather than a feature, with the orchestration layer described as ensuring that all actions and findings are explainable, reproducible and regulator ready, and the platform positioned around transparency and control in the compliance function. Reproducibility is the substantive part: an analyst or examiner can rerun a finding and obtain the same result, which is a stronger guarantee than an explanation generated after the fact.

The founding framing that agents are grounded in ledger data rather than reasoning freely constrains what the machine may assert. What is absent is the human boundary, with the process described as automated end to end and no review gate, escalation threshold or sign off point identified before a finding reaches a regulatory filing.

The verifiability property here is unusually strong and it comes from the subject matter. Findings are described as evidence backed and reproducible, with agents grounded in ledger data, and a blockchain is an immutable public record, so any claim about a transaction can be checked independently against a source the vendor does not control and cannot restate. That is the Daloopa property with a firmer foundation, since filings can be amended and ledgers cannot.

The company's framing that institutions receive decisions they can trust and defend is a claim about examinability rather than accuracy. What is missing is measurement of the inferential layer, where the real risk sits: no accuracy or false positive rate is published for risk scoring or attribution, which is where a reproducible process can still be reliably wrong.

Founded in 2024 and out of stealth only in October 2025, so there is no adoption record to assess. No customer is named, no transaction volume or chain coverage figure is published, and no outcome is measured. What exists is credential rather than evidence, and it is unusually specific: the founding team built the onchain data and financial crime infrastructure at one of the largest digital asset exchanges, operating at petabyte scale, with two further founders having led that exchange's crypto data engineering. A 15 million dollar seed was co led by two well known venture firms with participation from the venture arms of two major exchanges, which is strategic money from the customer adjacent side.

No data boundary statement was located, and the sensitive material here is not the ledger but the enquiry. In financial crime work the fact that a particular institution is investigating a particular address is itself confidential, since it can reveal a live suspicion, a customer relationship or an impending filing, and with multiple institutions on one platform two of them will inevitably examine the same address at the same time. Nothing states whether investigation activity is segregated, whether query patterns inform anything shared, or how the company treats the intelligence its customers generate through use.

No data protection agreement, retention schedule or subprocessor list was located. The privacy question in this category has an unusual shape: the source data is a public, permanent and pseudonymous ledger, so the vendor is not assembling a new store of private records, but the product's core act is attribution, linking addresses to entities and ultimately to people, and that inference is exactly what converts public pseudonymous activity into identified financial behaviour. Nothing published describes how attributions are formed, how confident they are, how long they persist, or what happens to a wrong one once it has been recorded.

No attestation, certification, trust centre or enumerated framework was located, which is expected for a company months out of stealth and is also the artifact that will gate its intended customers. Banks and government agencies do not onboard a financial crime vendor without an independently assessed control set, and the stated ambition to deepen partnerships with institutions and public sector bodies will run into that requirement before it runs into anything technical.

Regulatory readiness is the product's central claim and no regulation is named. Output is described as audit ready and regulator ready, and the company positions itself against a backdrop of authorities demanding that digital assets meet the same anti money laundering and counter terrorist financing standards as traditional finance, without identifying a single one of them.

For a product whose entire purpose is discharging financial crime obligations, the absence of the specific instruments, covering suspicious activity reporting, the rules governing transfers of digital assets between institutions, and sanctions screening duties, leaves the buyer to establish sufficiency alone.

No protected class decision applies, so the axis adapts, and two exposures replace it. The first is attribution error, since an address wrongly linked to an illicit entity taints not only its owner but everyone who has transacted with it, and in a public ledger that association is permanent and visible to every other analytics provider.

The second is risk propagation by proximity: transaction graph scoring means a person who received funds several hops from a flagged source can inherit risk they had no knowledge of and no ability to avoid, which in practice results in accounts frozen or refused. The consequence lands hardest on users in jurisdictions where digital assets substitute for banking access. No error rate, attribution confidence measure or correction mechanism was located.

No guarantee, indemnity or falsifiable accuracy commitment was located. The institution is well served on defensibility, since findings are described as evidence backed and reproducible and the company's own phrasing is that customers receive decisions they can trust and defend, which is the property that matters when an examiner asks why an account was frozen. The party with no route is the address holder.

Someone whose funds are flagged through proximity to a tainted address is not told which system reached that conclusion, cannot see the evidence chain, and has no described mechanism to contest an attribution that will persist on a public ledger indefinitely.

The primary data source is inherently disclosed, since public blockchains are the input and anyone can verify what they contain, which closes part of this question by subject matter. The rest is undisclosed and one omission is significant.

Risk scoring in this category depends on attribution datasets linking addresses to named entities, exchanges, sanctioned parties and illicit services, and those labels are proprietary, assembled and frequently licensed rather than derived from the ledger, yet no attribution source or partner is named. No model provider, hosting arrangement or subprocessor appears either.

Modularity is emphasised as an architectural principle and the platform is described as programmable, which points toward interface based consumption, but no integration is evidenced. No case management system, core banking platform, transaction monitoring suite or exchange system is named on the output side, no blockchain networks are named on the input side, and no developer documentation was located. For a compliance product the connection that matters most is into the institution's existing alerting and filing workflow, and nothing describes it.

No hosting provider, region selection, residency commitment or private deployment option was located. Exposure is lower than for vendors holding customer records, since the analysed data is public ledger activity, but institutions and public sector agencies conducting investigations will want to know where their enquiry data and case material rest, and nothing published addresses it.

No pricing, packaging or basis of charge was located, which is unsurprising at this stage. The modular architecture the company emphasises implies a component based commercial model, since an institution taking only screening has a different footprint from one running screening, investigation and reporting together, and nothing indicates whether charge falls per address screened, per transaction monitored, per chain supported or as a platform fee.

The intended buyer set is stated broadly, covering banks, fintechs, payment providers and public sector agencies, and multi chain support is a stated design goal with expanded network coverage funded by the seed round. None of it is evidenced. No institution type is demonstrated in production, no blockchain networks are named as currently supported, and the function is narrow by design, addressing digital asset compliance rather than the wider financial crime estate. Coverage here is an intention rather than a footprint.