Treffen Sie lyfX.ai unter Lubricant Expo Europe 2026

Discovery Theatre: AI co-creation for the concept phase of engineering

On 15 September 2026, André Moreira presents AutoGraph: AI-native concept engineering in the Discovery Theatre. Early-stage engineering is still manual work: concepts take weeks, assumptions are hard to trace, and knowledge stays buried in past projects.

AutoGraph puts AI agents that follow norms and specifications into this phase. PFD Bench turns a process flow diagram into an accurate process description in minutes, validated with K-UTEC Salt Technologies and open to try at pfdbench.lyfx.ai. The CAPEX Engine, co-developed with an industrial partner and in beta, produces early capital-cost estimates with transparent assumptions and calculations. The point is not to replace engineers, but to let them explore more innovative options earlier.

Download: AI Co-Creation

Real-world applications of AI in lubricants, with a case from ROWE

On 16 September 2026, André Moreira is one of the panelists in Real World Applications of AI in Lubricant Development & Formulation. His contribution, From AI Tools to Agentic Organizations, shows how a small AI-first company runs its daily work with AI agents, and used one real case as the example: a market deep dive, run by AI agents and built together with ROWE, one of Europe’s largest independent, family-run lubricant makers.

lyfX.ai ran the study as a human-AI team. A Claude Code orchestrator, running on Claude Opus 4.8, planned the work with the ROWE team, froze the application taxonomy and dispatched twelve specialist agents: five source extractors, a company profiler, four analysts (technology, market, competitors, industry structure), a report assembler and a monitoring drafter. People set the goals and made the calls; the agents prepared and executed. ROWE’s experts reviewed the result, and their feedback was used for further refinements of the results.

What came out

• A 72-page market study (22,000 words): application taxonomy, technology benchmark and SWOT per application, market sizing and forecasts, product compilation, competitor portfolios, value chain, Porter’s Five Forces and a customer list by region. Every claim cited.
• 12 agents, 2.18 million tokens, 2 h 20 min from kick-off to report, about 24 USD in estimated model cost.
• Three ready-to-run monitoring briefings, so the one-off study turns into ongoing market intelligence.
• The work continues: ROWE and lyfX.ai are working on further AI projects, including ongoing market intelligence.

Built with Claude Code and Claude Opus 4.8 (Anthropic) on FuzzyClaw, lyfX.ai’s open-source agent orchestration layer (Open Source with MIT licence).

Download: From AI Tools to Agentic Organizations