Skip to content

competitor-intel

Competitive analysis for EmbeddedEra as a contract electronics design firm. Private repo angelEmbeddedEra/competitor-intel.

Status

  • 2026-09-08: repo created. Profile, competitor-analyst agent, and rules in place. First landscape report in progress.

Decisions

  • Analyses services competition, not SaaS feature parity. The community competitive-intel skills (alirezarezvani/claude-skills, ai-sales-team-claude) all assume pricing pages, feature matrices, ACV and churn. A design shop competes on lifecycle coverage, minimum engagement size, and proof of past work instead.
  • Reports are dated and never overwritten, so the market's history stays readable.

Gotchas

  • The profile file is the baseline for every comparison. A stale profile silently produces a wrong report.

Findings that should not be relitigated

  • The defensible position is not "embedded design consultant". That lane is crowded above (Cardinal Peak ~1,000 engineers, Synapse ~150, Delve ~83) and below (offshore $15-70/hr, US freelance avg ~$103/hr), and PCB layout is being deflated by AI tooling (Quilter routed an 843-component 8-layer board).
  • The position is: "I build the instrument, the app that configures it, and the browser tool your customers use to update it, then I manufacture, test and ship it. One person, one contract, 50-5,000 units." Revised 2026-09-08 after the software layer was found; the earlier schematic-to-shipped-box wording undersold by about half and read as a manufacturing pitch.
  • ~~The full-lifecycle quadrant is empty.~~ This was wrong. Full-stack IoT agencies (Very, Softeq) occupy it, and Cardinal Peak does deliver web and app in-house. The quadrant is uncontested only at this engagement size and with physical custody of manufacturing. Do not claim "nobody at any size does this".
  • Golioth, Particle, Blues and Memfault are partly competitors, partly integration options — not purely substitutes as the first report framed them.
  • Web Serial does not work in Safari or on iOS. Disclose this rather than get caught.
  • Locally the bridge is unoccupied: Impact Embedded (Durham) stops at the prototype, GRT Electronics (Raleigh) starts at assembly. GRT is a partner, not a competitor.
  • Do not contest hourly rate against offshore. Unwinnable, and fighting it concedes the frame.
  • Bus factor is the most legitimate objection and must be answered in writing.

Plan

PLAN.md in the repo is the client-acquisition plan: five stations to PCB Carolina on Wed 11 Nov 2026. Unblock the site, say the one sentence, film the line, publish prices, then outbound to 30 instrument OEMs. Station 5 (outbound) is an addition; the reports only fixed positioning.

Log

  • 2026-09-08: Client acquisition plan written. Deadline anchor is PCB Carolina, 11 Nov 2026, attend-only this year (tech sessions already posted); talk is a 2027 target.
  • 2026-09-08: Addendum written. Corrected three conclusions from the morning's report. Verified the software layer is real: ScaleBuddy is ~17k lines of Dart with a full Firebase backend (Firestore, Cloud Functions, Google/Apple auth, an AI feature) paired with its own sensor hardware. Full span is sensor to shipped box with no subcontractor. Pacer is partial and 5k, Bike-Computer, Cow_Remote are empty scaffolds - never cite them as app work.
  • 2026-09-08: First landscape report written and verified; all 49 cited URLs resolve. Top actions: name the niche (industrial/scientific instrumentation OEMs), publish video proof of the SMT line, publish three fixed prices, sell the ATE station standalone, and refuse hourly staff-augmentation work.
  • 2026-09-08: Checked the community first per Hard Rule 1. Four maintained competitive-intel skills exist; all are SaaS-shaped. Borrowed their structure (competitor typing, positioning map, tracking cadence) and wrote a small adapted agent rather than vendoring a poor fit.