FAQ
General
Who are you?
A small team of security researchers. Our main practice is Web3 security: smart contract audits, pre-audit reviews, development and fuzzing. Our second, AI Systems, audits what companies spend on AI and builds small AI tools that do one job. See the Introduction.
How do we pay?
Each engagement has a fixed price, agreed before the work starts, and one of three payment terms: half at the start and half on delivery, all of it on delivery, or all of it up front (as for a grant's participation fee). We invoice in EUR or USD, and you can pay by bank transfer or in stablecoins.
How do we follow the work?
In the client portal: where the work stands, the offer, the findings or milestones, and the reports as they are ready. You sign in with your email address; there is no password.
Do you offer free work?
Through grants: cohorts with a fixed fund, for teams starting out. Applying is free; a small participation fee is due only if you are accepted. AI work also starts with a free first review.
Can I introduce a project to you?
Yes, through the referral programme: 20% of what the project pays, with no cap, for Web3 and AI work alike.
What is the badge?
A small, signed badge for your site that links to a page on ours saying what we certified: the Security Grade of an audit, a tested build, or a verified saving. See Project Badges.
Web3 Security
What do you audit?
Smart contracts and protocols written in Solidity and Rust, across EVM chains, HyperEVM and Solana (SVM). Our portfolio includes lending, AMMs, vaults, liquid staking, bridges, perpetuals, GameFi and real-world-asset protocols.
How does an audit run?
We scope the code and the threat model, run automated analysis and a line-by-line manual review, and share the findings as we go. Then a preliminary report, a review of your fixes, and a final report with your Security Grade. See Web3 Security Audits.
What is the free AI audit?
We run our own AI auditing tools over your repository and send you what they find, at no cost and with no strings attached. It finds a lot, but not everything a manual audit does. See Free AI Audit.
What is a pre-audit review?
A first pass on your code before the formal audit, so the audit spends its time on hard problems rather than obvious ones. See Pre-Audit Reviews.
How much does an audit cost, and how long does it take?
It depends on the size and complexity of the code. The price estimate gives a range in a minute, with the review of your fixes included; send us the repository and we confirm the price once we have read it.
Do you publish the reports?
Where the client agrees, yes. Our portfolio links to the full reports, and the final report is written to be publishable.
Do you compete in audit contests?
Yes, on Code4rena, Sherlock, CodeHawks and Cantina. The results are public.
AI Systems
Do you build chatbots?
Not as a product. We do build internal assistants that answer staff questions from your own documents, and we can draft replies inside your support tools. We do not sell generic chatbots.
Which models and providers do you use?
We choose based on the task, quality and cost. We work with the major providers. We will use the one you already have an account with where it makes sense, and we explain our reasoning.
Do you resell AI?
No. We do not resell subscriptions or API access. You pay providers directly, from your own accounts.
Is our data used to train anything?
No. We do not use client data to train anything. We work with your own accounts and API keys where possible. Provider terms on data retention differ, so we check them with you. Details are agreed in writing for each engagement.
What if the AI is wrong?
It sometimes will be. That is why tools have human review where mistakes are costly, review queues, failure handling, logging and a way to switch them off. See Data, Access and Review.
Do you work with small teams?
Yes. We are a small team ourselves. Many good first workflows come from teams of a few people who spend hours on repetitive tasks.
How long does a first workflow take?
Typically a few weeks. It depends on scope, access and how quickly we get examples. We do not guarantee a timeline before scoping.
Can you reduce our bill without lowering quality?
Often some waste, such as duplicate calls and retry loops, can be removed with no effect on output. Other changes carry quality risk, so we check with an evaluation set before any model swap. We do not guarantee a percentage. Savings are measured against your invoice.
Do you need production access?
Not for a review. For a bill review, read-only billing and usage access is usually enough. If building or changing something requires more, we ask for the least access needed and agree it in writing.
Who owns the code?
The work we build for you is meant to run in your accounts, with a named owner on your side. Ownership and licensing terms are set in the written agreement for each engagement.
What happens after handover?
The tool is yours and runs on your accounts. The first 30 days are covered while we measure the result. After that you can look after it yourselves, or take a care plan: Care at €350 a month or Care Plus at €900 a month. See Care Plans.
How do I start?
Ask for the free first review. See Working With Us.