Decision Intelligence

Every Tool Helps You Find Past Decisions. This One Helps You Make Better Ones.

Your team makes hundreds of decisions a month. Six months later, most are invisible - buried in Slack threads, lost with departed colleagues, relitigated from scratch. Decision Intelligence captures decisions in 60 seconds, challenges them with AI before they're locked in, and makes them findable forever. Built on human-AI partnership research and a decision log running continuously since November 2025.

Book Discovery Call See How It Works

The Knowledge Problem Every Growing Company Has

You've tried wikis. You've tried documentation sprints. You've tried recording meetings. None of it stuck. Here's why.

  • Confluence/Notion creates a second place to look. People default to Slack. The wiki decays within 90 days.
  • Documentation sprints produce overhead nobody maintains under deadline pressure.
  • Meeting recordings are 45 minutes nobody rewatches. The decision is in minute 37.
  • AI search over garbage returns garbage faster. The input quality problem is upstream.
  • The real issue isn't storage. It's capture discipline - making it effortless to log the decision at the moment it's made.
The Knowledge Problem Every Growing Company Has

Four-Layer Architecture

Capture decisions in 60 seconds. Store them in a knowledge graph. Challenge them with AI before locking in. Find them instantly when needed.

Layer 1: Capture

A Decision Card triggered by /decision in Slack. Eight fields, 60 seconds. Records not just what was decided, but what was rejected and why - including explicit alternatives (Option A/B/C), the second choice, confidence score, and expiry date. The most valuable field: 'what we almost decided instead.'

Layer 2: Memory (Semantic Knowledge Graph)

Decisions become nodes in a semantic graph with typed relationships: supersedes, depends_on, contradicts, relates_to. People and teams are nodes. When someone leaves, their decisions remain. A visual timeline shows decisions as filterable nodes. Deduplication is built in - the system searches before writing.

Layer 3: Decision Quality (the layer nobody else has built)

AI Adversary Test generates steel-manned counter-arguments. Cognitive Bias Scan detects 55 bias patterns. Management 3.0 Delegation Poker surfaces authority misalignment. Red Team rotation mandates structural dissent. Concern Round converts silence into explicit consent. PARDES Five-Reader Engine for the highest-stakes decisions.

Layer 4: Retrieval

AI-native search with three modes: direct query ('what did we decide about billing?'), contextual suggestion (surfaces prior decisions when a new Slack thread starts on the same topic), and onboarding digest (new hire gets their domain's decision history on day one).

What Makes This Different From 'Just Use Notion'

Every knowledge management tool on the market is an archive with a search bar. This system has a challenge layer - it makes decisions better at the moment they're being made, not just findable after the fact.

  • AI Adversary Test: not 'are you sure?' but a specific, steel-manned counter-argument based on your stated alternatives
  • 55 Cognitive Bias Detectors: organized by decision type (pricing, hiring, product, strategy, technology) - each a specific question, not a label
  • 63 Thinking Tools: Inversion, Second-Order Thinking, Circle of Competence, Margin of Safety - AI selects 2-3 relevant models per decision
  • PARDES Five-Reader Engine: five analytical lenses for the decisions that shape the company - produces emergent insight none of the lenses show alone
  • Management 3.0 Delegation Poker: cards revealed simultaneously to surface authority misalignment without anchoring
  • Red Team Protocol (US Army doctrine): structural dissent as a rotating assignment, not a personality trait
  • Concern Round (Quaker + Sociocratic + Special Ops AAR): converts silence from ambiguous to explicit
  • Two tiers: everyday decisions get the AI check (90 seconds). High-stakes decisions get the full human + AI toolkit (30-60 minutes)
What Makes This Different From 'Just Use Notion'

Built on Proven Patterns

This system wasn't designed in a whiteboard session. It emerged from daily use.

  • 90+ decisions logged with explicit alternatives, confidence scores, and second-choice tracking - continuously since November 2025
  • Semantic knowledge base with cross-session search: 130,000+ indexed chunks across four collections, multilingual embeddings, deduplication by design
  • Cognitive bias library tested across consulting, research, and product development decisions
  • Protocols drawn from Management 3.0, US Army Red Team Handbook, Quaker decision-making, sociocratic consent governance, and special operations after-action review
  • Full methodology published and open for inspection: RAZEM framework (maciejjankowski.com/methodologies/razem/), Zbigniew Protocol (GitHub)
  • The complete architecture is public: Full System Design (18 min), the executive summary, and the deck
  • The Open Window: the research-backed protocol layer, including the cognitive-science evidence the system has to survive (14 min)
  • Try the thinking before you buy the system: four free AI prompt tools - run one real decision through the Decision Board, no signup
  • The 'alternatives + second choice' card design emerged from a specific discovery: the most relitigated decisions were the ones where nobody recorded why Option B was rejected
Built on Proven Patterns

Sealed Pre-Mortem: The Per-Decision Way In

Not ready for a system? Bring one decision. A facilitated pre-mortem before a major decision is locked in - assumptions stated, probabilities assigned, earliest failure signals named. Then the record is sealed, timestamped, and opened at the 60-day review. Your decision process becomes auditable separately from its outcome.

The Session: 2-3 Hours, One Decision

A facilitated pre-mortem before the decision is made. The team states its assumptions explicitly, assigns a probability to each, and names the earliest failure signals - observable events, defined before the outcome is known, not after.

The Seal

The record is timestamped and content-addressed (an IPFS pin): cryptographic proof of what was written and when, without disclosing the content to anyone. At the 60-day review we open it together and score every assumption against what actually happened.

Why Seal It: Outcome Bias Protection

Six months later, decisions get judged by how they turned out, not by how they were made. A sealed pre-mortem gives the decision-maker a provable record that the risks were named in advance - the process becomes defensible even when the outcome goes against you.

The Price: 1,900 EUR Fixed

One session, one decision, sealed record and 60-day review included. Fully creditable against a full Decision Intelligence engagement.

Start With a Question You Can't Get Answered

Bring your hardest knowledge management problem. We'll show you what a Decision Card would look like for your most recent relitigated decision - and what the AI adversary would have said about it.

Book a Free 15-min Call