Caverio trader companion: overnight conceptual-design programme (started 2026-09-26 00:50 Beirut)

Source: two Thomas voice notes, 00:08 and 00:28 on 2026-09-26, in the Vesper direct lane. Vesper is co-founder on this and runs the programme. Bolo is the build counterpart and is not yet engaged.

The decision (Thomas, 00:28, his words condensed)

• The Now / Situation / People ladder is out. What follows is not a v2. It is the build.
• Everything built before tonight was experimentation and is to be labelled as such.
• Tonight is the moment of conceptual design: the product is now clearly identified.
• First step is to plan everything: an infrastructure lane (system, sources, what to gather, how to organise data, what to capture and not capture, efficiency in tokens, bandwidth, CPU, RAM) and a product-design lane in steps: curated experience skeleton, then wireframes (the logic of the interface: what you see where, a chart of this, a big number of that), matched against the data infrastructure and adding to it where needed, then per component the presentation mechanics that make the presentation proprietary, then visual design, then connecting modules.
• He expects most of it to build in a few days, then production, then onboarding. Start with beta testers: a feedback-for-access programme (one month free for feedback; useful feedback extends a month; none or bad feedback drops you).
• Overnight order: think through everything said, break it into topics and tackle them one at a time, so that by morning there is an immense amount of research-backed and data-backed material on everything described, how to do it, and what to show. Tomorrow he cleans it up and building starts.

The product (Thomas, 00:08, condensed)

A trader companion: one view that pulls every source a trader would otherwise open in five apps, so a buy / no-buy / exit / size decision can be made fast without switching apps. The system gathers data that is hard and costly for one person to gather, and presents it so that it is useful. It does not judge; it shows. Three layers:

1. Follow the money (market brief). State of crypto as an environment: money flowing in and out, stablecoins held as firepower, how that has historically moved markets; then money by chain and platform (which chain is hot, terminal volumes as a degen-activity read, attention flow of traders); then BTC versus alts, top 50 versus low caps, which market-cap band is hottest (money in relative to size), today / week / month views for different trader speeds. This is the brief on the environment you trade in.
2. Trading activity (discovery into situations). Start with meme coins; later other market-cap categories that behave differently (for example a $20M to $50M band with legroom for slower traders, narratives per niche, social watch on narrative heat). Discovery is an explore list, curated but not tight, fed by: tokens mentioned on X by solid accounts; tokens mentioned by Telegram callers (expand the caller list; read whole chats, not just formatted calls; track each caller's stance per token and detect bullish to bearish flips); tokens bought by interesting on-chain wallets; tokens bought by Fomo-profiled traders. A look list, not a buy list. Every discovered token opens a situation; as more happens it lights up in categories and rises. Not the old ladder; a different design. In the situation: market cap, price, volume, money flow, social watch (sentiment direction, intensity, mentions relative to peers in the same band). Curated lens: less to look at than the whole market, more to look at per token. Presentation is the proprietary part: which fields, in what form (chart, gauge, number, sentence, diagram), studied deeply.
3. Strategy zones (later, harder). Per strategy, buy zone / hold zone / sell zone highlights. Win is defined forgivingly: did the token move up enough after the trigger, for long enough, that a person could have taken the trade and won. Trade management stays with the trader; the situation room must be telling enough in the hold stage. No entry/exit strategy product for now.

Topics and workers

Phase 1, nine parallel workers, briefs in briefs/, artifacts named there, target 90 to 150 minutes each:

Phase 2, one synthesis worker launched automatically when all nine artifacts end in Status: DONE:

| T10 | Synthesis: decisions per topic, contradictions, module list, honest estimate, questions for Thomas | MASTER.md + INDEX.md |

Boundaries

• Everything here is planning and research. No production code changes, no service restarts, no writes outside this directory. Data reads are read-only and go through copies or read-only sqlite connections.
• The 09-25 business plan (../business-plan-2026-09-25/PLAN-v3.md) is not cancelled, but its 2 Oct record-page and 9 Oct paid-open dates were set for the ladder. T9 says what survives; Thomas decides tomorrow.
• Nothing is sent to anyone outside. Bolo is briefed tomorrow from MASTER.md.

Rules

See RULES.md.