Straight answers.
Every one traceable.
Practical questions about token standards, forward-deployed AI engineering, and running AI agents inside regulated finance. Each answer links to the page or post that carries the full working.
// TOKEN STANDARDS & TOKENIZATION
What is ERC-3643 (T-REX)?
ERC-3643, or T-REX, is the Final Ethereum standard for permissioned security tokens. It adds recipient identity checks, configured transfer rules and issuer controls. Privileged operations have their own rules. It supplies engineering mechanisms; it does not establish that an issuance meets every legal requirement. I am one of its five named authors.
The full ERC-3643 explainerShould I use ERC-3643 or ERC-1400?
Start with the instrument and its permission model. ERC-3643 specifies on-chain identity and token controls. The ERC-1400 family includes partitioned balances and controller operations. Both need implementation choices and requirements from the legal and compliance team; the standard name alone does not decide the fit.
ERC-3643 vs ERC-1400 vs CMTAT, comparedHow much value is tokenized on ERC-3643?
The ERC3643 Association reports more than $32B in assets tokenized using the standard. This is the association's ecosystem figure, not value I personally tokenized and not a live valuation of assets still outstanding. The explainer links to its source.
What ERC-3643 is and where it is usedWhat is ERC-6960?
ERC-6960 is a Draft token interface with two identifiers. A mainId identifies an asset category and a subId identifies a class or variation within it. Balances record amounts of each pair, so a share class can contain many fungible units. I am its first author, one of six.
The full ERC-6960 explainerWhat is ERC-7410?
ERC-7410 is a small ERC-20 extension. It lets a spender lower or revoke the allowance an owner granted it, without waiting for the owner to submit that change on-chain. It cannot be used to increase an allowance. It is a Draft, and I am listed second among its three authors.
The full ERC-7410 explainerHow does an idea become an Ethereum standard?
A proposal follows the EIP process: specification, editorial review, community discussion and status changes. Receiving a number or being merged into the repository does not make it Final. My practical lesson is to make the interface, assumptions and tradeoffs clear enough for other implementers to review.
The anatomy of an ERC, start to finish// FORWARD-DEPLOYED AI ENGINEERING
What is a forward-deployed AI engineer?
A forward-deployed AI engineer builds and ships AI systems inside the customer's environment, against their real data and their real constraints, instead of handing over a design and leaving. In regulated finance that changes the job: the audit and compliance requirements are part of the engineering problem from day one rather than a review at the end.
A practitioner definition of the FDE roleWhat does an AI agent engineer actually do?
An AI agent engineer builds the software around model-driven workflows: tools, integrations, permissions, evaluations and operations. My July 24, 2026 snapshot contains 119 AI/agent-engineering postings from 32 employers; 81.5% mention an agent framework or pattern. That selected sample describes posting language, not a definitive boundary for the profession.
The State of Agent Engineering, with the dataset// AI INSIDE REGULATED FINANCE
Can an AI agent be trusted to move money?
Start with the permitted action, not the model's confidence. My approach is to keep payment preparation separate from approval and execution, enforce narrow tool permissions and record what was authorized. The responsible team must define who may approve, what happens on an uncertain result and how repeated requests avoid duplicate actions.
Least-privilege agents that move moneyHow do you stop a RAG system from hallucinating in finance?
There is no universal guarantee that a RAG system will never hallucinate. I combine exact source references, faithfulness checks, numeric verification and abstention when evidence is insufficient. Model-based checks can also fail, so they need evaluation cases and human review appropriate to the consequence of a wrong answer.
Verified citations and abstention in RAGWhat does the EU AI Act require from an engineering team?
This is engineering guidance, not legal advice, and your obligations depend on your system and your counsel. In practice the parts that reach an engineering team are about evidence: logging what the system did, showing that a human can oversee and override it, and being able to demonstrate accuracy rather than assert it. Those have to be built in. Retrofitting them is the expensive path.
The EU AI Act as an engineering checklistHow do you know an AI agent is ready to ship?
Agree on acceptance cases for the workflow, including forbidden actions, missing evidence and integration failures. Use evaluations to block regressions, then review monitoring, rollback and the handover owner. A passing test suite is evidence for the defined scope, not proof that every production risk is covered.
The eval suite that gates a release// WHO IS ANSWERING
Who is Adam Boudjemaa?
I am an AI and agent engineer for regulated finance, based in France, and a former CTO. I am a named author on four Ethereum standards, three of them merged (ERC-3643, ERC-6960, ERC-7410) and ERC-8203 still a draft in review, and I have shipped production multi-agent, RAG and LLM systems inside regulated organizations. The long version, with dates and receipts, is on the about page.
About, with the full recordLooking for scope, pricing or availability?
Those are different questions and they have their own answers: what an ERC-3643 audit covers, what I currently have capacity for, how engagements are structured, and what happens if it does not work out.
▸ Read the services FAQQuestion not here?
Tell me what you are building and where you need help.
▶ ASK MEPrefer email? contact@adam-boudjemaa.com