Three layers. All live. The AI-native business is complete.
Every production AI capability at Lifetime Oy is organized into three layers that build on each other. Agents are the unit of work. Loops give them the ability to self-correct. Workflows chain them into end-to-end outcomes. All three are live as of June 2026.
50 production Skilled Agents, live in src/agents/. Every agent extends
BaseAgent and emits a ledger-native output contract on every execution:
journalEntries (immutable audit trail),
regulatoryArtifact (machine-readable compliance evidence), and
lineage (full provenance chain). Firehorse 7 standard throughout.
Loop Engineering landed in PR #1447 (commit d34876a),
8 June 2026. A Maker-Checker cycle: one Skilled Agent
proposes; a second validates; the loop continues until the outcome is correct or the
budget stops it. State is disk-persisted — loops survive restarts. Two gates guard
every cycle: a budget gate (cost ceiling) and a
safety gate
(KYA trust score).
src/agents/loop-engineering/ · PR #1447 · commit d34876a
· Maker-Checker · disk-persisted · resumable
Multi-agent orchestration via the /workflow skill and
src/agents/workflow/ orchestrators. Three proven patterns:
fan-out (parallel agents on subtasks),
sequential (ordered pipeline), and reflection (critic
loop). MCP Hub integration enables cross-tool fan-out. Live use cases: EU compliance
assessment pipelines, Smart Factory build chains, KYA governance workflows.
src/agents/workflow/ · MCP Hub · fan-out / sequential /
reflection · live
The fleet is organized into three tiers. Orchestrators (Skilled Agents that coordinate) run at the top. Industrial Specialist Agents (per-industry domain experts) sit in the middle. Horizontal and expansion agents handle cross-cutting capabilities. All run in EU-sovereign infrastructure; all emit the Firehorse 7 ledger-native output contract.
Platform-served (no dedicated specialist agent): Manufacturing (cross-sector), Healthcare, Financial Services, Public Sector, Data Centres.
CSRD (Corporate Sustainability Reporting Directive), CBAM (Carbon Border Adjustment Mechanism), ETS (EU Emissions Trading System), and the EU AI Act are not future-tense. They are in force. Every EU company with over 250 employees must comply. That is more than 50,000 companies by 2028. These are not dashboards. They are not reports. They are not consulting hours. They are production Skilled Agents that observe, think, and act.
Lifetime Oy is a System-of-Action operator, not a SaaS vendor. We ship outcomes — compliance artifacts, audit trails, regulatory evidence — not software licenses.
Double materiality assessments, ESRS disclosures, supply chain sustainability data. Athena (Compliance) runs the pipeline. Ledger-native output, audit-ready.
Embedded carbon declarations for imports and exports. Industry-specific emission factors. Talos handles Aluminium; Vulcan handles Steel; Gaia handles Cement.
Allowance tracking, verified emission reporting, surrender obligations. Continuous monitoring, not annual scrambles. Helios (Energy) leads; sector agents contribute.
Article 12 requires a full audit trail for high-risk AI systems. Every Skilled Agent emits a signed, hash-chained journal via the Auditor agent. Compliance is built in, not bolted on.
KYA (Know Your Agent) is Lifetime Oy's runtime agent governance layer. Every Skilled Agent is subject to KYA controls before, during, and after execution. No agent acts without a verified trust score; no loop runs past its budget gate.
Trust score validation. Each agent carries a KYA trust score. Low-trust agents operate in a restricted sandbox. Score degradation triggers automatic escalation to a human operator before the next cycle.
Budget constraints. Every Loop has a cost ceiling enforced by the budget gate. When the ceiling is approached, the loop suspends and emits a BLOCKED signal. A human approves continuation or terminates.
TC-4 Leash Snap. Chain termination protocol TC-4 can halt any agent chain immediately — propagating a STOP signal through every active Workflow. Designed for adversarial or runaway conditions. No bypasses in production.
EU AI Act Article 12 audit trail. The Auditor agent writes a hash-chained journal entry for every agent action. Board-readable. CFO-auditable. Admissible evidence for regulatory inspections.
All production Skilled Agents run on European cloud infrastructure. Data never leaves European jurisdiction. US providers are not used for production workloads. Aegis (DWS IQ's dual-use platform) is EU-only by design.
EU Infrastructure Partners
We amplify human intelligence in regulated industries.
Three layers. All live. Outcomes, not dashboards.
Lifetime Oy is a System-of-Action operator for European industry, where compliance can freeze or free capital. 50 production Skilled Agents. Loop Engineering. Multi-agent Workflows. EU-sovereign infrastructure. Available now.