SENTINEL MAN is not a single AI responding to prompts. It is a coordinated system — four specialised engines working from your data to produce intelligence that improves the longer you use it.
Each engine has a distinct job. Each draws from the same data store — your check-ins, your decisions, your Operating Principles — and produces a different form of intelligence. None of them work in isolation. Together they build a model of how you operate that no single AI session can produce.
Every morning, before you open your inbox, SENTINEL MAN produces your operational brief. Not a generic motivational message. A data-driven intelligence report, specific to your state on that specific day.
The brief begins with your mode assignment — COMMAND, RECON, or STANDBY — determined by the system from your check-in data and recent trend. Your mode sets the tone for everything that follows. A COMMAND mode brief is aggressive and directive. A RECON brief is observational, diagnostic. STANDBY is minimal — the system knows today is not the day to push hard.
Below the mode: your Readiness Score (0–100), your decision quality window (the hours your data says your judgment is sharpest), and three operational levers — specific, actionable moves for that day. No generic advice. Each lever is derived from the pattern the system has built about you.
Above the daily call sits your weekly objective — one headline objective per Monday-week that you set on COCKPIT. SENTINEL reads it back to you every morning until it is done or you mark it breached. One call per week. Not a list. The discipline is forcing one call.
Five inputs. Under 90 seconds. No typing unless you choose to. The check-in is deliberately minimal — because a tool you do not use daily produces no intelligence at all.
The five inputs — mood vector, sleep hours, physical load, mental clarity, and network pulse — feed directly into your Readiness Score calculation. A sixth optional field, carry-forward, lets you log an open item in under ten words. That item appears in tomorrow's brief until you clear it.
From Day 7 onward, your scores are calculated against your personal baseline — not population averages. A Readiness Score of 71 means something different for you than for the man next to you. The system learns the difference.
Three modes. COUNCIL — five AI lanes synthesised in 26 seconds. FAST — sub-5-second risk flag for live negotiations. REPLAY — past decision retroactively assessed against the data that was available at the time.
Every council assessment ships with calibration disclosure, multi-OP conflict surfacing, black-swan tail risks, and alliance-risk forecasting when the decision involves a contact in your network. The council ends with a SOVEREIGNTY CHECK — a question, not an answer.
Reasoner — checks the decision against your history and your Operating Principles. Has this pattern appeared before? What did your data look like on days you made similar choices?
Contrarian — attacks the obvious choice. Specific kill shots. Worst-case outcome. What you're not seeing because you're inside the decision.
Context Enhancer — second and third-order consequences. Relational ripples. People adjacent to the decision who aren't in the room.
Predictor — what you'll actually do. Genuinely evaluating, or seeking validation for a call you've already made?
Pre-mortem — most likely 90-day failure mode + the early-warning signal that would tell you it's failing in time + one guardrail worth attaching before you commit.
COUNCIL — full five-lane assessment. 7 per day. For decisions where 26 seconds of analysis earns its keep.
FAST — sub-5-second Haiku response. One risk flag, one question. 10 per day. For live negotiations, board meetings in progress, deals closing on the phone.
REPLAY — submit a past decision plus what actually happened. The council retroactively assesses what it would have said at the time. 5 per day. Empirical proof that gut + council beats gut alone — using your own history.
Calibration disclosure — a collapsible card showing exactly what data the council was reasoning from. Operator record counts, cohort archetype matched, lanes that ran. Trust by disclosure, not by claim.
Multi-OP conflict resolution — when two of your active principles fight each other (e.g. "move fast" vs "72-hour buffer"), the system names the tension and hands the resolution back to you.
Alliance Risk Forecast — when the decision text mentions a contact in your network, an estimate of relational damage probability over the next 90 days, with the specific dynamic named.
Black-swan tail risks — 1–2 low-probability high-impact scenarios with the early signal that would tell you they're activating.
The Decision Engine is capped at 7 council calls per day (FAST and REPLAY have separate quotas). The cap forces selection — you decide which decisions deserve the full council. That selection discipline is itself part of the product.
The council is not your echo chamber. From Day 1, before any personal pattern data exists, the lanes draw from 1,600 simulated decisions across 8 archetypes matched to your profile. The Contrarian runs on a different model architecture with no access to your principles. External reference is built in.
Every call you log outcomes against makes the next call sharper. After 50 logged outcomes the Reasoner knows your hit rate by decision type, which conditions correlate with your best and worst calls, and which lane was most predictive. Compounding intelligence over time.
Most tools show you how you felt today. SENTINEL MAN builds a model of how you operate over time — detecting the correlations that repeat in your data before you're consciously aware of them.
Two scans (daily light + weekly deep), three operator-facing outputs (Cohort Percentile, Network Activation, Dynamic Principle Refactoring). All grounded in the patterns the system has confirmed in your data — never population averages.
The Pattern Engine runs two scans. A daily light scan checks for emerging signals — possible correlations in your most recent data. It never surfaces a pattern until the evidence threshold is met. A weekly deep scan runs a full analysis across your history, confirming patterns, updating confidence scores, and retiring patterns that no longer hold.
The patterns that matter most reveal your decision conditions — the physiological and situational states that predict your best and worst calls. "Decision quality drops 31% when sleep debt exceeds 45 minutes and physical load is high." That's not a population statistic. That's your data.
Your decision-quality and readiness benchmarked against the synthetic operator cohort matched to your archetype (founder under cash stress, executive in transition, late-career professional, etc.). Percentile rank, mode-mix comparison, plus one top-decile gap tactic.
Activates with 10+ rated decisions or 14+ daily logs in the last 90 days. Solves the solo-operator's lack of external calibration without exposing personal data to peers.
When your contacts list shows overdue people, SENTINEL generates 1–3 short outreach drafts prioritised by leverage to your weekly objective, with a recommended channel (text / email / DM / call). Operator-voice, ready to copy-edit-send.
3 generations per day. Reframes the network as a tactical asset for the week, not a guilt list.
Every 7 days at most, SENTINEL reads your decision outcomes + principle adherence + active patterns and proposes ≤3 changes to your Operating Principles (additions, edits, retirements). Each cites specific decisions as evidence.
You accept what fits. Reject what doesn't. The system never edits your principles for you. The accumulation moat made operative — your foundational rules iterate from your own data, not from a template.
Every Sunday evening, a performance review on your week — not a mood summary. The arc continues quarterly (THE SIGNAL) and annually (Day 365 review). Searchable archive becomes your intelligence record on yourself.
Plus an exportable Performance Ledger — credentialed, dated, call-sign-stamped. The kind of record you walk into a Series A, a board review, or a new engagement holding.
Sunday evening. Not a mood summary. A performance review — the kind a serious athlete would review between training sessions.
Covers what changed, what degraded, what improved, what your decision quality looked like, and which Operating Principle was tested during the week. Closes with your primary objective for the next cycle — a single actionable focus that carries forward automatically to Monday's brief.
On the last day of every calendar quarter (Mar / Jun / Sep / Dec), THE SIGNAL fires — a 280–400 word digest of what crystallised, what is drifting, and what the next quarter demands.
After reading THE SIGNAL you know whether the trajectory of the last 90 days is the trajectory you intended. The 90-day reflection most operators promise themselves and never do — written from your record, not a template.
At checkin_count === 365, a five-section verdict closes the year-long arc: the operator you were on Day 1, the operator you are now, what the record proves, what it warns, and the next year's call.
The longest-arc artefact in the product. Most performance/AI tools never accumulate enough history to write a meaningful annual; SENTINEL does because the daily check-in is the ground-truth substrate.
The DEBRIEF tab is searchable. Every brief — weekly, milestone, quarterly, annual — has clickable section anchors and a master/detail reader on desktop. Search across the full archive with hit highlighting. By Week 12 your debriefs are an intelligence record on yourself.
The Performance Ledger is exportable as a credentialed, dated, call-sign-stamped record. A founder going into a Series A. An executive going into a board review. A consultant pitching a new engagement. "Here is my decision record — what I was right about, what I was wrong about, what the pattern shows." No other product produces this.
Every check-in, every decision, every principle edit, every weekly debrief feeds the same long memory. After 180 days of use the system holds a confirmed pattern library of how you operate, a decision history of every significant call you've made with outcomes attached, the lifecycle of every principle you've created or retired, and a calibration baseline tuned to your specific archetype.
The longer you use it, the more expensive it becomes to leave. That's not a threat — it's the value. No competitor can give you back six months of your own data.
Every tool claims to be different. Here is the specific difference.
Seven days. No credit card.
The system starts building from Day 1.