System trace

4 weak signals → hypothesis

Four signals over three weeks. Each one alone: inconclusive. Together, with temporal context from git, they form a directed arc — and a high-confidence hypothesis with a concrete next move.

18-day observation window 1 synthesis pass · 1 human decision

Scenario

Company: Ax##ra — fictional B2B SaaS, ~120 engineers, post-Series C.
Period: three weeks of passive observation. No outreach, no direct contact.
Output: a dated micro-hypothesis and a typed outreach action with channel logic tied to the signal type.
Synthetic names. Real signal taxonomy, real confidence rules, real temporal logic.

The signal arc

Four signals accumulate. Confidence builds with each — and with their sequence.

Week 1 · Feb 3

leadership-change moderate

New CTO announced at Ax##ra.

LinkedIn announcement from T**as W**vik on his first day. No stated agenda yet.

id: S-31 · company: Ax##ra · signal_type: leadership-change · strength: moderate
date: 2026-02-03 · source: linkedin-post · linked_person: T**as W**vik

New CTO at Ax##ra. 90-day window open.
Strength: moderate — announcement only, no stated agenda yet.
One signal. Too early for a hypothesis.

System state: signal filed. No action triggered. Confidence too low for a hypothesis on one signal alone. A watch window opens for Ax##ra.

↓ 8 days later

Week 2 · Feb 11

tool-search weak

CTO comments on a Kubernetes migration post.

"We're rethinking our deployment model — interested in what's worked at scale."

id: S-35 · company: Ax##ra · signal_type: tool-search · strength: weak
date: 2026-02-11 · source: linkedin-comment · related_signal: S-31

CTO engaging publicly with infra tooling content.
Strength: weak — a comment, not a decision signal.
Context: 8 days after joining. Platform decisions likely still unmade.

System state: two signals. Same person, same window. Pattern forming — not yet confirmed. related_signal: S-31 links them explicitly in the file.

↓ 3 days later

Week 2 · Feb 14

hiring-surge moderate

Three backend roles posted in 5 days.

Senior Backend Engineer (×2), Platform Engineer — posted 11 days after the new CTO started.

id: S-37 · company: Ax##ra · signal_type: hiring-surge · strength: moderate
date: 2026-02-14 · source: vacancy-board · related_signals: [S-31, S-35]

Three backend roles posted 11 days after new CTO start.
Hiring pattern consistent with: new leader staffing up before a platform change.
Strength: moderate — vacancies are public intent, not confirmed budget.

System state: three signals. Two weeks. Same company, converging direction. market-analyst will pick these up on the next scheduled run.

↓ 7 days later

Week 3 · Feb 21

public-complaint strong

CTO names the problem publicly.

"Three weeks in. The hardest part isn't the technology. It's the delivery process that nobody has written down."

id: S-42 · company: Ax##ra · signal_type: public-complaint · strength: strong
date: 2026-02-21 · source: linkedin-post · related_signals: [S-31, S-35, S-37]

CTO names delivery process legibility as the primary blocker.
Public, named, specific. Strength: strong.
Window: same 90-day new-leader period. Problem statement is out in the open.

System state: four signals. 18-day arc. The last one names the exact problem. market-analyst now has enough to synthesize with high confidence.

↓

↺

Synthesis — market-analyst reads the arc

Four signals become one directed pattern.

market-analyst sees signals as a sequence, not a snapshot. Git timestamps are part of the data — the 18-day arc is visible.

# market-analyst reads the timeline

Signal timeline for Ax##ra:
  2026-02-03  S-31  leadership-change  moderate  → new CTO starts
  2026-02-11  S-35  tool-search        weak      → platform thinking begins
  2026-02-14  S-37  hiring-surge       moderate  → staffing up
  2026-02-21  S-42  public-complaint   strong    → names the exact problem

Pattern: new CTO, 18-day arc, escalating specificity.
The last signal names the exact problem: undocumented delivery process.
Direction: consistent. Confidence: high.

Skill invoked: market-analyst

↓

Hypothesis — dated, confidence-scored

A micro-hypothesis with confidence: high.

# market-state/hypotheses/ax##ra-delivery-legibility.md

---
id: H-29 · level: micro · confidence: high
date: 2026-02-21 · signals: [S-31, S-35, S-37, S-42] · expires: 2026-05-21
---

T**as W**vik (CTO, Ax##ra) is in the 90-day new-leader window.
Named problem: delivery process that "nobody has written down."
Hiring to fix it. Thinking about platform. Talking about it publicly.

Narrow, time-bound window. High openness to outside perspective.
The problem is not AI — it's the delivery legibility that blocks AI adoption.

Confidence: high
Why: 4 signals, 3-week arc, strong final signal naming the exact problem.
     Sequence matters: the pattern is directional, not coincidental.

↓

Action — channel tied to signal type

Signal S-42 was a public post. The first touch is a public comment.

# market-state/actions/AI-31.md

---
id: AI-31 · kind: outreach · hypothesis: ax##ra-delivery-legibility
who: T**as W**vik, CTO, Ax##ra · channel: LinkedIn comment (on S-42 post)
---

Angle: his post named the problem exactly.
Respond to the post — not with a pitch, but with a question that shows
I've seen this problem and have a specific frame for it.

Message draft (comment):
  "That gap between how delivery actually works and what's written down
   is the exact thing that makes AI agents fail in engineering orgs.
   Not the model — the undocumented decision tree.
   How far along are you in mapping it?"

CTA: one question. No pitch. Respond to what he named, not what I want to sell.
KPI: any reply or follow-up within 7 days.
Stop: silence → log in market-state/people/wernvik.md, no retry for 30 days.

Channel logic: public signal (post) → public first touch (comment). A cold DM on a public post looks like a scrape. A relevant comment on the exact post that named the problem looks like genuine attention.

What this trace produced

signals over 18 days

synthesis pass

high

confidence · H-29

AI-31

comment · ready to send

No single signal was sufficient. Confidence came from accumulation and sequence — not from one strong data point.

Architecture decisions

Why not act on the first signal?

Every new CTO is a leadership-change signal. Acting on all of them produces noise. The system waits for accumulation. The first signal opens a watch window — not an outreach brief.

Why does sequence matter?

Four signals on the same day would be a cluster. Four signals over 18 days is a trajectory. The system uses git timestamps to reconstruct the arc. A directed pattern has higher confidence than the same signals without dates.

Why comment, not cold DM?

Signal S-42 was a public post. The system matches channel to signal type. Commenting on the exact post that named the problem is contextually natural. A DM that references a public post looks like a scrape.