Battle-card coverage of the AEC AI landscape

Coverage at a glance

Bottom line (2026-05-09, post fill-in): 43 cards on real AEC companies + 5 thought-experiment cards = 48 total. In-scope landscape coverage (Design + Preconstruction + Construction + Horizontal) is now 39 / 40 = 97.5% — the only gap is OpenSpace as a standalone, since it's bundled into the Buildots+OpenSpace combined card. Every in-scope segment is fully covered. Permitting and Operations/FM remain intentionally out of scope; Planon stays as the lone Operations card because it's a paper-anchor for Evaluator Power. The cast is now complete enough to run the AEC software war game across the entire AI landscape; the remaining work is the matchup-balance pass.

By segment

Permitting codes, compliance, approvals

Design generative, MEP, review, BIM

Preconstruction estimating, scheduling, coordination

Construction monitoring, field ops, safety

Operations / FM facility mgmt, digital twin, energy

Horizontal cross-phase

Outside the standard ConTech segment grid

What this round added (2026-05-09)

15 starter-sketch cards across the in-scope segments (Design, Preconstruction, Construction, Horizontal). All drafted at the same fidelity — moat type primary/secondary/tertiary, stat block, ai_era flags, 2–3 strategies, special move with trigger and effect, valuation-input rough order of magnitude. Matchups omitted on this batch — pending the H1 weighting fix on the existing Snaptrude vs Autodesk balance.

Tier 1 (5)

• Forma — Autodesk-internal AI design; primary data_flywheel, secondary switching, tertiary scale. Includes the Rhumbix-folded-in field-data feedback loop.
• CoLab Software — design-review collaboration on 3D models; primary network, secondary switching. Multi-user review network as the primary moat.
• ALICE Technologies — megaproject schedule optimization; primary data_flywheel, secondary switching. Megaproject-schedule corpus as the privileged-data thesis.
• Doxel — CV progress vs. BIM in data-center / lab construction; primary data_flywheel, secondary cornered_resource. Narrower ICP than Buildots/OpenSpace.
• Briq — 200+ financial-automation bots in GC finance; primary switching, secondary agentic_lock_in. The Loud Claim 2 multiplicative-composition exemplar.

Tier 2 (10)

• Augmenta, Endra AI, Hypar, Bild AI — AI-native design / MEP startups; counter_positioning + data_flywheel profiles vs Autodesk's seat-licensing model.
• XBuild, Togal.ai, CONXAI — Preconstruction estimating/takeoff/agentic-workflow attackers.
• Dusty Robotics, Fyld — Construction hardware/safety with cornered-resource and evaluator-power components respectively.
• Nomic AI — cross-phase document AI; the cornered-resource codes corpus is the differentiator vs single-phase competitors.

What's deferred

Out-of-scope for the current war game (still in the landscape but not getting cards):

• Permitting (4 companies): PermitFlow, UpCodes, CodeComply.AI, CivCheck.
• Operations / FM (7 companies): BuildOps, Willow, Entouch, Infraspeak, Facilio, Nantum AI, Mapped. Planon stays in (paper-anchor for Evaluator Power) but the rest of the segment is deferred.

Within scope but not yet carded:

• Design: Transcend ($33M, water/power infra generative), Archistar ($23M, AU site planning), Maket ($4M CAD, gen-AI floor plans), BuildCheck AI ($5.9M, CV error detection), Semble AI / Structured AI / Ark Design AI / Infraspace / AnkerDB (early/YC).
• Preconstruction: Rebar ($14M, MEP quoting), Brickanta ($8M, precon AI OS), Nyfty.ai (RFI/submittal bots).
• Construction: Sensera Systems ($27M, OSHA detection), Foresight (~$10M, megaproject PM), Moab ($16M, ops AI).

These are good candidates for a Tier 3 fill once the H1 weighting and matchup balance are stable. The current 24-card cast is enough to play the (7+4) war game across Design, Preconstruction, and Construction without leaning on the co-AEC thought experiments to fill missing archetypes.

Known balance issue

Snaptrude attacking Autodesk reads as Snaptrude losing too much ground in H1. Likely culprits to inspect together: the H1 horizon weights in type_chart.json over-favor distribution_reach + brand_power + capital_reserves (all stats Autodesk dominates and Snaptrude doesn't); Snaptrude's special move only fires against switching-primary defenders (Autodesk qualifies, but the +40% Round 0 may be small relative to the Round 0 weight); Autodesk's special move (“API perimeter fortification”) gives +30% defense at H1 with no opposing trigger condition. Recommend: review H1 weights, raise Snaptrude's H1-relevant stats, or add an attacker-side counter-trigger on Snaptrude's special. The new cards on this batch all stamp last_updated: 2026-05-09 for clean diffing once the rebalance lands.