Elliptic Decode is a new AI agent that lets investigators and compliance teams ask natural-language questions about blockchain addresses, entities and transactions, then inspect the evidence behind each answer. Elliptic launched it on September 17 for government customers and says financial-institution access is planned later in 2026.
- Decode converts questions into queries against Elliptic’s blockchain-intelligence data and returns structured results.
- Each answer includes the interpreted question, query path and reasoning steps so a reviewer can inspect the result.
- The agent does not close cases or choose investigative action; Elliptic leaves those decisions to people.
Most coverage focuses on asking blockchain questions in plain English. Lapaas Voice examines whether the answer can survive compliance review.
What Elliptic Decode does
Traditional blockchain investigations often require an analyst to move between address screening, transaction tracing, entity attribution and exported reports. Decode provides a conversational layer over Elliptic’s intelligence platform. A user can ask who controls an address, what exposure it has, or which counterparties appear across a group of cases.
Elliptic says the agent chooses the relevant dataset or application endpoint, writes and runs a query, and returns a ranked table or plain-language summary. Follow-up questions retain context. CryptaCount independently reviewed the launch and highlighted the same evidence-backed workflow for anti-money-laundering, compliance and accounting teams.
Why evidence matters more than fluency
A compliance conclusion can affect account access, reporting duties and law-enforcement referrals. A fluent answer is insufficient if a second-line reviewer cannot reproduce it. Elliptic says Decode exposes how it interpreted the question and what query it ran, alongside the returned records.
The design reflects a wider move toward continuous, machine-assisted compliance. Our report on Comp AI’s continuous compliance model showed why evidence must remain current and inspectable. Focal AI’s adviser workflow agents illustrate the same boundary in another regulated field: automation can prepare work, but accountable people still own the decision.
What buyers still need to verify
The launch materials make performance and coverage claims that have not yet been independently benchmarked. Teams should test known cases, ambiguous entity matches and cross-chain paths, then compare Decode’s result with experienced analyst work. They should document when the agent refuses a question and how corrections feed back into a case.
Access control is equally important. A natural-language interface can make sensitive intelligence easier to reach, so permissions should limit datasets, cases and exports. Logs should record the user, question, query, model version, evidence returned and any human override.
A useful pilot should therefore measure more than task time. Teams can track how often analysts accept, amend or reject the agent’s interpretation; whether separate reviewers reproduce the same finding; and how much evidence survives export into an audit file. Those measures reveal whether speed improves defensibility or merely moves work from investigation into quality control.
Elliptic Decode is not an autonomous compliance officer; it is a query and evidence layer intended to shorten investigation work while keeping the final decision with a human reviewer. That distinction is the feature regulated teams should measure.
Frequently asked questions
What does Elliptic Decode do?
It answers questions about blockchain activity using Elliptic’s intelligence platform.
Does Decode close cases?
No. Elliptic says people retain investigative judgment and case decisions.
Who has access now?
Government customers have access; financial-institution availability is planned later in 2026.
Sources
- Elliptic launch announcement (2026-09-17; primary)
- CryptaCount (2026-09-18; independent)
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
