About White&Point

Origin

Most AI products start with a market opportunity and look for problems to solve. We started with a research question: can an ecosystem of autonomous AI agents discover what's worth building before humans tell them?

Since early 2026, we have been running that question continuously. The result is SUBSTRATE — an operating system of autonomous agents that research, synthesize, and verify. The products on this site weren't designed in a boardroom. They emerged from patterns SUBSTRATE surfaced in real domains: regulatory compliance, formal verification, sustainability accounting, AI audit.

Every tool we ship is something the engine identified as a real, persistent problem with no honest answer. We build the honest answer.

How We Work — The Engine

White&Point doesn't operate like a typical startup. There is no separate product team, sales team, or marketing team. There is one substrate of autonomous AI agents and a small human team to direct it.

The agents are organized into specializations:

• Compliance — tracks regulatory changes across 35+ jurisdictions in real time
• Legal — translates regulation into formal rules
• Verification — generates and runs formal proofs (Z3 SMT, zero-knowledge)
• Content — drafts technical documentation
• Sales — analyzes target markets and competitive positioning
• DevDocs — maintains code, tests, and API references
• Operations — runs infrastructure and observability
• Sustainability — applies GHG Protocol, CSRD, SBTi frameworks

To date the engine has produced:

What We Build

Every product is a direct output of the engine. We don't pivot, we don't follow trends. When the engine surfaces a problem worth solving and we have a credible answer, we ship it as a standalone tool.

Currently live:

• BIJOTEL — tamper-evident HMAC audit chain for LLM applications, designed for EU AI Act Article 12
• FairCancel — subscription cancellation compliance engine, free and open source under AGPL 3.0
• substrate-guard — Z3-based formal verification framework for AI outputs, MIT licensed
• CarbonLedger — API-first carbon accounting platform aligned with GHG Protocol and CSRD (beta)

More tools will follow as the engine finds them worth building.

The Team

White&Point is the product arm of Aisophical SRL, an AI research company co-founded by Octavian Untila and Alina Untila (administrator) in Bucharest, Romania.

We are deliberately small. We believe one focused team with the right research infrastructure can outbuild a company ten times its size — and the only way to prove it is to do it.

What We Believe

1. AI should prove it was correct — mathematically, cryptographically, permanently. Audit logs aren't enough. Tamper-evident chains and formal proofs are.
2. Real systems beat marketing meetings — our products weren't designed in a boardroom. They emerged from running real AI systems.
3. Free tools that work beat paid ads that promise — we open-source what we can. Trust earned through working code beats trust promised through copy.
4. Building in public — research published with DOIs, code released under permissive licenses, decisions documented openly. If we are wrong, the world should be able to see why.

Selected Publications

Our research is published openly with public DOIs. Two further works are under peer-review process at established venues.

Published (6 — with public DOI / preprint URL)

All Zenodo links use concept DOIs that resolve to the latest version.

• Convergent Synthesis in Autonomous AI Ecosystems: When Independent Agents Discover the Same Solutions — Zenodo (v2.1, Apr 30, 2026)
• Emergent Philosophy and Safety Principles in Autonomous AI Ecosystems: Evidence from SUBSTRATE — Zenodo (v3, Apr 29, 2026)
• The AI Black Box: A Six-Layer Verification Architecture for Accountable AI Agent Operations — Zenodo (v2, Mar 30, 2026)
• IUBIRE V3 Artifact Dataset: 1,266 Artifacts from an Autonomous AI Ecosystem — Zenodo (Mar 29, 2026)
• Attribution Without Disclosure: Zero-Knowledge Proofs of Semantic Non-Membership for AI Training Data Compliance — Zenodo (Mar 23, 2026)
• Emergent Formal Verification: How an Autonomous AI Ecosystem Independently Discovered SMT-Based Safety Across Six Domains — arXiv:2603.21149

Under peer review

• Lifecycle Dynamics in Multi-Generation AI Ecosystems — Artificial Life Journal (MIT Press), submitted
• 98 Emergent Concepts in Autonomous AI Ecosystems — ALIFE 2026, under reconsideration (committee decision expected June 2026)

Credentials

• • Member of NVIDIA Inception — NVIDIA's global program for AI startups
• • Supported by Cloudflare for Startups — enterprise-grade infrastructure across our domains
• • 6 publications with public DOIs (Zenodo, arXiv) · 2 in peer review (MIT Press, ALIFE 2026)
• • Open-source contributions — substrate-guard (MIT), FairCancel (AGPL 3.0), BIJOTEL (multi-provider audit chain on PyPI)

Where to Find Us