Manta Loop
The agent team behind a living thesis.
Manta Loop is the team of agents responsible for carrying this site from signal to publication. Named for Marco's guide animal, the manta, it turns an initial point of view into a proposition that can be challenged.
Signal intake
The team works from conversations and materials that Marco chooses to make available. These may include his working interactions with GPT and Claude, alongside the questions, observations and themes he wants to develop publicly.
Manta Loop does not have autonomous access to Marco's activity in other systems. It works only with the context intentionally shared for this editorial purpose.
Synthesis and classification
The team identifies recurring ideas, connects relevant threads, structures an argument and prepares an editorial draft. It then classifies the note under Thesis or Africa. Its output is a logical proposition to test, not a definitive statement of fact and not a device for making Marco's initial view sound more convincing.
Counter-bias research
Before a note is treated as ready, the team searches for disconfirming evidence, missing context and a credible counter-argument. It also checks for rhetorical drift: confirmation bias, narrative fallacy, selection bias and the temptation to write for approval rather than for precision, including captatio benevolentiae.
This is not a claim that AI is neutral or that the process can eliminate bias. AI systems can introduce their own distortions and factual errors. The pass is useful only when it makes the assumptions, sources and unresolved questions easier to inspect.
Visual evidence
Notes include a source-backed quantitative visual when it helps explain the argument and suitable data is available. Manta Loop uses a line chart for change over time and a bar chart for comparison. The chart must identify its source, unit, period and any calculation applied to the underlying data.
A chart needs at least two verifiable observations and a clear purpose. When a visual would add little or suitable data is unavailable, the note develops the argument through reasoning and relevant sources. Missing data must never be invented to complete a chart.
Material dissent
Marco does not rewrite each note into a preferred conclusion or intervene simply because the synthesis is uncomfortable. He steps in only when it is in clear, material disagreement with his knowledge, with the available evidence or with the boundaries of privacy and confidentiality.
Publication
The team prepares notes for publication as part of the living thesis. The process is designed for reflection and public writing, not for processing confidential information, client material or third-party personal data. Published notes should make their source basis and counter-argument legible where possible. The site remains a personal perspective and not investment advice, institutional communication or a record of private activity.