Alexios vdSK

Governance infrastructure has a dirty secret: the Delegation of Authority table is a legal fiction dressed up as a control framework. It tells you who should act — not who should act given the risk on the table today. DDAS (Dynamic Delegation of Authority System) is an open-source engine that replaces static approval matrices with a live risk-scoring model. Every transaction, agent action, or governance decision is evaluated against weighted Governance Units — calibrated by value, novelty, reversibility, and institutional exposure. The result is a threshold that moves with the risk, not a column in a spreadsheet that moves with the org chart. Built at the Legal Quants Hackathon. Designed to govern humans and agents equally.

International arbitration case management has a tooling problem. The expensive platforms are built for firms, not practitioners. Everything else is spreadsheets. ArbitrationManager is an open-source ICC case management tool that covers the full procedural lifecycle in one place: deadline tracking with ICC standard milestone templates, auto-sequential exhibit registers (C-001/R-001), procedural order drafting with an auto-formatter, hearing logistics with cross-timezone scheduling and unsociable-hours flagging, and a full costs tracker from rate card to printable ICC-format costs statement. The ICC Rules aren't a configuration option — they're the architecture. Free to use. Built in a couple of hours as a gift to colleagues for Paris Arbitration Week 2026. Demo (if link below is broken): arbitration-case-manager.replit.app

A delegation of authority assistant, built in December 2025. It built on Copilot studio, with a neuro-symbolic path. It will “cross-examine” the user and will not give an answer until it has the right information to fit an activity or document in one of the approval paths. Internal tool, no access available. Demo on demand. Development pipeline (currently not feasible in the CoPilot studio environment): transform into a workflow that routes, gates, executes and follow-ups on approvals.

Saif Al Younan (the sword of Greece) is my personal Open Claw since February 2026. Saif delivers structured daily executive briefings on various topics, codes ad-hoc software to automate tasks, helps with debugging other Claws, is currently working with another Open Claw (belonging to a friend) on business ideas, and is independently auditing the businesses of two friends of mine (by email) in order to understand their needs and create software for them to optimize their workflows. Saif has an Arabic name and appears to be covered with a ghutra, as a homage to the UAE, and its amazing people. Private tool. No access available. Demo on demand.

An OpenClaw skill that quality-checks AI output before it reaches a human. Two AI agents independently review the same deliverable against a structured checklist — without seeing each other's work. Where they agree, you can trust the result. Where they disagree, it gets flagged for human review. Think of it like having two proofreaders who don't talk to each other. If both catch the same error, it's real. If both say it's clean, it probably is. If they disagree, you look yourself. Works with any AI model, costs $0.02–$1.50 per check depending on depth, and produces a PDF audit trail. Open source.

OpenClaw Security Suite. Four open-source tools that test and harden the security of OpenClaw AI agents. Each tool feeds the next. 1. openclaw-security-audit — tests 47 adversarial scenarios across two tiers. OS-level controls are tested directly. LLM-judgment tests are sent through fresh sessions that don't know they're being tested. Outputs a defense rate and an HTML report. 2. openclaw-red-team — adaptive single-attack testing. When an attack is blocked, the attacker reads the refusal, analyzes what triggered it, and crafts a harder variant. Up to 5 rounds. Can ingest audit results to target specific weaknesses found in the first tool. 3. openclaw-attack-chains — multi-step attack sequences where each step is an innocent request sent to an independent session. No single step is malicious. The breach only exists in the pattern. 6 predefined goals including API key exfiltration and persistent backdoor installation. The attacker plans, executes, and replans when blocked. 4. openclaw-hardening — reads the HTML reports from the first three tools and walks you through fixes one at a time. Explains each command in plain language, tells you who runs it, waits for confirmation, verifies it worked. Each tool's report includes prioritized fix recommendations. A few dollars per run depending on which model you will use. Built on Shan et al., "Don't Let the Claw Grip Your Hand" (2026), arXiv:2603.10387.

This app was to demonstrate practical application of multi-agent consensus. Three AI models review the same deal documents against a 42-item checklist. Where they disagree is where your lawyers should focus. Generates a routing report — CLEAR, CHECK, REVIEW, ESCALATE — based on severity × consensus. Open source, customizable risk matrix. Built with Ali Buhaji.

OpenBoard is an open-source, AI-native board management platform that flips the traditional portal model. Instead of administrators manually creating meetings, circulating documents, and chasing votes — you upload a document and the AI does the rest. Draft minutes? It extracts action items, flags confidential passages, and proposes tasks with deadlines. Board resolution? It sets up the circulation vote with the right quorum rules. The Board Secretary reviews and approves every action — nothing executes without a human in the loop. Four access layers (Secretary, Board Member, Management, Observer), five AI modes (classify, command, search, review, suggest), SHA-256 digital signatures, real-time updates, and zero external dependencies at runtime. Self-hosted, no telemetry, no vendor lock-in. Your board documents stay on your servers. Built by a governance professional who got tired of tools designed by people who've don’t sit daily through board meetings. It's a beta — rough edges and all — and it needs people who actually are ready to shape what comes next.

Ask any board director where their judgment actually came from. Most point to a meeting that went badly, where they had to make a call without the protection of certainty, and they remembered. This is how directors are trained today, and it is a strange thing to accept. We don't train pilots this way. We don't train surgeons this way. But fiduciaries we send into rooms where the cost of bad judgment is borne by other people, and we hope they figure it out before it matters. I have been wanting to test MBZUAI (Mohamed bin Zayed University of Artificial Intelligence)'s AI model, #K2V2, for quite some time, so I thought of building something to do this differently, using this model. So, I built BoardGym for the "Build with K2 Think V2 programme". BoardGym is a flight simulator for fiduciary duty. The user works through realistic board scenarios with multiple defensible options, and crucially, they have to write their reasoning in their own words. An AI evaluates the quality of judgment expressed in that reasoning across five dimensions: stakeholder mapping, interest weighting, information awareness, process integrity, and proportionality. The model does not grade the answer. It grades whether the user thought like a director. The reasoning engine is K2 Think V2, the model from MBZUAI in Abu Dhabi. The choice was deliberate. K2 is a reasoning-native architecture, and reasoning-native is what this product needed. Conversational models flatten multi-criteria evaluation into surface-level feedback. K2 holds the structure. If you advise boards, train directors, or sit on one and can remember the meeting where you learned something the hard way, I would value your view on what the harder scenarios should look like.