LLM-Powered Tools
Language models are a capability layer, not a strategy.
Marcelo builds practical tools around OpenAI, Claude and Gemini APIs for marketing and commercial work, with the value coming from domain judgement and workflow design.
Where they help
- Content generation within defined brand and factual constraints.
- Document analysis, extraction and executive summarisation.
- Classification and routing of marketing or commercial inputs.
- Research assistance and structured comparison.
- Internal productivity tools that reduce repetitive knowledge work.
Where they do not replace judgement
Language models can produce confident errors, over-generalise and optimise for plausible language rather than commercial truth. The system design should therefore include source control, human review and explicit boundaries around what can be published or acted upon automatically.
Capability status: documented source materials describe these tools as built and used in Marcelo’s own operation. They should not be presented as a large portfolio of client production deployments until those cases exist.