/company-intel

Company Intel — Executive Signal Refresh Example

All companies, executives, quotes, figures, and URLs are fictional, using the same surrogate universe as the SKILL.md examples. This shows the rerun pattern: the user previously ran company-intel on Helix Motion Systems (the baseline excerpts appear in SKILL.md's Examples section, dated June 2026), and now asks again a quarter later. The agent has the prior output in context, so it defaults to delta-first reporting — no "refresh" keyword needed.

Trigger: "Run company-intel on Helix Motion Systems again — earnings were last week."

What's Changed Since June 2026

1. The digital twin program has a name, a P&L owner — and a smaller scope

• Then: volume hiring for digital-twin engineers and IoT platform architects across the Diversified Industrial segment; inference was that hiring ran ahead of organizational readiness.
• Now: the program surfaced on the Q3 call as "Helix Foresight," assigned to the Aftermarket Services unit — "Foresight is an aftermarket offering first; we'll expand from our installed base rather than chase greenfield" (CFO, Q3 earnings call, Oct 2026, transcript URL).
• Reading: the platform ambition got fenced into the services P&L — the segment-based org structure won, as the baseline inferred it would. For competitors: greenfield industrial-IoT deals just got less contested — Inference.
• Confidence: high (named program, named owner, direct quote)

2. Hydrogen language shifted from products to partnerships

• Then: patent clustering in hydrogen fuel-cell components; R&D outpacing product announcements — read as a decarbonization bet.
• Now: two co-development agreements with OEMs announced; the CEO described hydrogen as "a partnered path to market" (investor day, Sep 2026, URL). Patent filings in the class continue — Fact (filings index).
• Reading: the bet continues but the go-to-market changed — component supplier, not system seller. The competitive threat model shifts from "new rival product line" to "their components inside our OEMs' machines" — Inference.
• Confidence: medium (two data points, one quarter)

Dropped Language

• "Segment digital scorecards" — mentioned on four consecutive earnings calls through Q2; absent from the Q3 script and slides entirely. Paired with the Foresight re-scoping, the internal digital-metrics push looks quietly retired — a transformation program that died without a press release.
• "Platform" — appeared 11 times in the Q2 materials, twice in Q3 (both times about Foresight). The vocabulary contraction matches the scope contraction; the narrative shift preceded any org announcement, which is why earnings materials get read as strategy documents.

Section 9 delta (Strategic Signals)

• Leadership: the VP of Digital Transformation (hired 2025 from a platform-first rival) departed in September — four months after the strategy she embodied was re-scoped — Fact (announcement URL). Consistent with the "senior leader exits within 6 months of a strategy shift" HUMINT chain: the strategy is in trouble, or already re-decided.
• Hiring: digital-twin postings down from ~40 to 6, now all within Aftermarket Services — Fact (job boards, Oct 2026).
• Patents: hydrogen filings steady; no new digital-platform classifications this quarter.

Sections with no material change

Sections 2 (revenue model), 3 (customers), 5 (product lines), 7 (competitors): no significant change since June 2026. Second consecutive quiet quarter for Section 5 — for a company this acquisitive, sustained product-line quiet is itself worth one watchlist line.

Why this example works

• It diffs; it doesn't regenerate. Eleven sections existed in the baseline; the refresh leads with two material shifts, a Dropped Language read, and one delta-focused section — plus an honest "no change" list. The reader spends three minutes and knows exactly what moved.
• Dropped Language carries the headline. Nothing Helix said announced the digital retrenchment — the scorecard vocabulary vanishing and "platform" contracting 11→2 told the story a quarter before the VP's departure confirmed it. What leaders stop saying is the signal.
• The Then/Now form keeps the baseline honest. Each shift quotes what the prior run recorded, so the inference chain (June: "hiring ahead of readiness" → October: "the org structure won") is auditable rather than retconned.
• Quiet got graded, not padded. Four sections report no change in one line each — and one sustained quiet is flagged as a signal of its own.