Gradon, real and false C2 nodes, and ACE expeditionary sites

Question: Can Gradon play a role in building C2 nodes and false C2 nodes, and in making ACE expeditionary sites? Which WWII concepts come back, and what new ones come from Ukraine and the Indo-Pacific?

Short answer: Yes, in one narrow role. Gradon could be the drawing and calculation engine inside the engineering firms that design Pacific airfields, dispersed sites and C2 facilities. That work is mostly 1391s, 35% designs, site and civil packages, and adapting standard designs.

It cannot be the C2 or deception provider. Today it also cannot touch controlled military work at all: it is a two-person, cloud-backed commercial product with no CUI, CMMC, FedRAMP, UFC or on-premise story.

The opening is real. The Pacific Deterrence Initiative (PDI) carries about $2.7B a year of construction. Deception now has a growing Air Force budget line. Nobody has shown generative AI producing DoD design packages.

1. What Gradon is

Gradon is the trading name of Aisle Technologies, Inc., a Delaware company in San Francisco. It is in Y Combinator’s Fall 2026 batch, YC lists its team size as two, and it is backed by YC and Founders, Inc. (YC, terms). The founders are Nour Zahzah (CEO, Berkeley mechanical engineering, former mechanical design engineer) and Yousef Helal (CTO, Berkeley, Stanford MS, FPGA work at Apple and AMD).

What it does:

• A desktop agent that drives the firm’s own installed software. Named tools include Civil 3D, AutoCAD, Revit, MicroStation, OpenRoads, STAAD, ETABS, HEC-RAS, HydroCAD, ArcGIS and Bluebeam.
• Engineers message it work like a teammate. It first ingests the firm’s past projects and standards into a “Firm Brain,” which the company stores in its own cloud (heygradon.com).
• Sample outputs on its site: a 62-sheet civil construction set in Civil 3D, and a 25-sheet structural set with a 191-page calculation package.
• Licensed engineers review everything. The terms say Gradon “does not seal or sign any work.”

Claims and the gaps behind them:

• Claims: project time cut by more than 75% (one launch example went from 750 hours to 180), “2x revenue per engineer,” and forward-deployed onboarding of at least a week on site.
• No named customers and no press.
• No defense presence: the site never mentions defense, UFC, ITAR, CMMC, FedRAMP, SOC 2 or air-gapped deployment.
• Unfinished privacy page: it is marked “DRAFT, PLACEHOLDER TEXT, NOT YET IN EFFECT” (privacy).

2. The precedents

WWII: deception was a construction program

• UK decoy department. Colonel John Turner’s department built daytime dummy airfields (K sites), night lighting decoys (Q sites) and, after Coventry, Starfish fire decoys. By war’s end there were 237 decoys protecting 81 towns and cities (Starfish).
• Fortitude South. It combined dummy landing craft built from scaffold, canvas and oil drums with radio traffic scripted from recordings of real exercises. The traffic simulated eleven divisions of a fictional army group, and German belief in it held into August 1944 (English Heritage).
• The Ghost Army. The 23rd Headquarters Special Troops had about 1,100 men and ran 22 operations (Ghost Army). Its units covered three channels:
• Visual: engineer camouflage with inflatables.
• Radio: a special signal company sending false traffic.
• Sound: a sonic deception company.
• Soviet maskirovka before Bagration. Dummy armor, false radio traffic and radio silence led Hitler to move more than half of Army Group Centre’s tanks and artillery away from the real axis (IWM).
• Repair and build speed.
• Henderson Field crater fills took about 40 minutes per raid (Seabee history).
• Tinian North Field reached four runways of about 8,000 ft and hardstands for 265 B-29s within ten months (North Field)).

Throwbacks that apply again:

• Decoy airfields (K and Q sites), now as decoy ACE spokes.
• Lighting and fire decoys at night.
• Dispersed hardstands and revetments.
• Pierced steel planking reborn as AM-2 matting. A 96 by 96 ft Marine pad takes about a day after survey and grading (3rd MAW).
• Rapid crater repair.
• Scripted signals deception built from real traffic.
• Engineer camouflage units that build to match real use, including vehicle tracks and wear.

Russia and Ukraine: decoys work only when they match the real signature

• Painted bombers failed. Painted Tu-95 silhouettes at Engels were obvious in commercial imagery (TWZ).
• Three-dimensional decoys worked.
• Ukrainian wooden HIMARS decoys drew Kalibr strikes (TWZ).
• Metinvest builds repairable metal-frame artillery and radar decoys (Kyiv Post).
• Decoys now guard air defenses. Russia places mock Pantsir and Tor systems around real ones (Defence Blog).
• C2 exposure. Ukraine struck Russian headquarters more than 22 times in eight months of 2022 (Military Review).
• Russia adapted. It pulled major HQs beyond rocket range into hardened structures and linked subordinates by directional microwave to cut emissions (RUSI).
• RUSI’s 2022 rule. HQ survival comes from dispersal, mobility or hardening; a civilian building alone does not protect it (RUSI).
• Operation Spiderweb. On June 1, 2025, truck-borne FPV drones hit bombers on open aprons deep inside Russia (Forecast International).
• Russia then built shelters. The UK MoD confirmed shelter construction at about 14 airbases (Army Recognition).

New concepts from Ukraine:

• Signature-matched 3D decoys.
• Decoys clustered around high-value emitters.
• Emission discipline as the real C2 survival factor. That means remoted antennas, directional links and decoy emitters physically separated from staffs.
• Light overhead cover against FPV drones, as well as full hardened shelters.
• Control of commercial satellite comms. Starlink whitelisting in February 2026 cut off Russian-used terminals (NBC Palm Springs).

PLA and the Indo-Pacific: China hardens and deceives at scale

Shelters. Within 1,000 nm of the Taiwan Strait, the PLA has 134 air bases with more than 650 hardened and about 2,000 non-hardened shelters. Over the same period the U.S. added 2 hardened and 41 non-hardened shelters in that radius (Hudson).

Decoys and mock-ups:

• J-20-like shapes appeared at Lintao in 2022 (TWZ).
• Full-scale U.S. carrier and destroyer outlines sit on Taklamakan ranges (WIFR).
• PLA doctrine stresses inflatable and solid decoys plus false communications (The Diplomat).

U.S. exposure:

• A House China Committee letter noted no hardened shelters on Guam or in the CNMI. It cited a CSIS wargame finding that 90% of U.S. aircraft losses occurred on the ground (House letter).
• Air Force leaders still favor dispersal over hardening (TWZ).

Taiwan:

• Taiwan plans 26 more hardened shelters for 2027 to 2031 (Taipei Times).
• Stimson models Taiwan’s runways closed for the first two to three weeks of a conflict (Stimson).

Implication: shelters multiply an attacker’s aimpoints, and decoys multiply them again. Hardening and deception work together; neither replaces the other.

ACE and EABO today

Doctrine: AFDN 1-21 moves air operations “from centralized physical infrastructures to a network of smaller, dispersed locations” (CRS).

Pacific site work:

• Tinian North Field: RED HORSE rehabilitated more than 20 million square feet of WWII pavement (Air & Space Forces).
• Yap: about $400M planned (Air & Space Forces).
• Peleliu: Marine engineers rebuilt a 6,000 ft strip.
• Philippines: nine EDCA sites.
• RAAF Tindal: U.S.-funded B-52 aprons.

The FY2027 PDI book says ACE needs “concealment, hardening, pre-positioned essential supplies” (FY27 PDI).

3. The money

Line	FY2025	FY2026	FY2027 request
PDI total	$8.72B	$10.06B	$11.72B
PDI construction (MILCON)	about $1.66B	about $2.77B	about $2.74B
Navy MILCON design (planning and design money)			$144.9M
Minor construction (Navy and Marine Corps $134.1M, Army $76.3M)			$210.4M
Air Force Tactical Deception, PE 0208007F (all of it the Air Base Resiliency project)	$0.5M	$13.5M	$48.9M
Navy Naval Construction Forces, PE 0603239N (airfield damage repair, advanced base construction)	$7.6M	$13.1M	$7.7M

Notes on the table:

• Tactical Deception. The FY2027 jump funds air base defense prototypes. It also absorbs “Airbase Kits,” moved from an electronic warfare line.
• Naval Construction Forces. FY2026 includes a $4.8M congressional add for “Autonomy Kits.”
• PDI construction figures are sums of the PDI book’s Table 4.

C2 construction in the FY2027 PDI:

• Joint Consolidated Communications Center, Andersen: $132.4M.
• MDA Guam Defense System Command Center: $99.7M.
• Joint Communication Upgrade, Guam: $60M.

Decoys have no procurement line. They show up as small buys:

• Misawa inflatable radar and infrared decoys, about $552K.
• A Marine Corps inflatable HIMARS and radar decoy solicitation.

4. Who buys the design work

Projects start as paperwork. MILCON projects begin as DD Form 1391 justifications. They are designed to 35% before budget lock, and planning and design funds pay for it (ERDC Code 2, 10 U.S.C. 2807). Congress now imposes design deadlines on congressionally added projects.

The construction agents:

• USACE Honolulu, Japan and Far East districts.
• NAVFAC Pacific and NAVFAC Marianas.
• AFCEC.
• RED HORSE and the Seabees, who design organically from standard designs. Those standards are moving to the Joint Construction Management System, which delivers AutoCAD templates (AFIMSC.pdf)).

Design is done by A-E firms on IDIQ contract vehicles:

• NAVFAC Indo-Pacific A-E, $990M.
• AFCEC Guam A-E MATOC, $930M, small business.
• USACE Japan, $249M (Woolpert-RS&H).
• USACE Korea, $200M (AECOM, HDR, Jacobs, Tetra Tech and others).
• AFCEC AE Next.

The two $15B NAVFAC Pacific construction vehicles are design-build, so their designers of record are A-E subcontractors.

Closest precedents for AI in this work:

• AFRL/AFWERX SBIRs for “Automated Propagation of Airfield Design for RED HORSE” (FA864922P0964) and for AI contingency airfield site selection (FA864924P1066).
• AFCEC’s stated push toward AI agents for faster infrastructure delivery and cost estimating (AFCEC).

No public source shows generative AI producing 1391s, 35% designs or full plan sets for DoD.

5. Where Gradon could fit, and where it cannot

Real fits (analysis)

1. Design sprints for dispersed sites. ACE needs many small sites, each needing:
2. survey and geotech base drawings;
3. grading and tropical drainage;
4. pavement thickness design to UFC 3-260-02;
5. AM-2 mat pads;
6. fuel bladder and hydrant points;
7. small power and water systems.

This is volume civil work, which is exactly what Gradon claims to compress. The constraint the Air Force faces is the number of sites, not the size of any one site.

1. 1391 and 35% packages. These include parametric cost estimates. Pacific design money ($144.9M of Navy design in FY2027 PDI alone) runs against congressional design deadlines.

1. Adapting standard designs. Gradon could take JCMS and AFCS templates in AutoCAD and fit them to specific Pacific sites.

1. Decoy sites designed with the same rigor as real ones. WWII and Ukraine both show that a decoy fails when its layout, tracks, utilities, lighting or thermal pattern differ from the real thing. A tool that produces a full real-site package and a matching decoy package from the same template makes “one real site, several false ones” cheap at the drawing level.

1. C2 node facilities. Gradon could design the shell, power, cooling and antenna remoting, and the setback layouts for small dispersed C2 posts and their decoy twins. Emitter decoys, signals deception and the C2 systems themselves belong to signals and electronic warfare firms, not to a design tool.

Blockers

• Data and security. Facility drawings for C2 and decoy sites are CUI at minimum, and the layout itself is the secret. Today Gradon:
• stores each firm’s “Firm Brain” in its own commercial cloud;
• promises only “commercially reasonable” security;
• has an unfinished privacy page.
• Compliance. CMMC Level 2 third-party certification is now a condition of award for CUI work. A cloud tool holding CUI needs FedRAMP Moderate equivalency, and government-hosted use implies Impact Level 4 or 5. C2 and decoy work will usually need an on-premise or air-gapped build.
• Criteria. UFC is never mentioned. Outputs would have to meet UFC 3-260-01 and 3-260-02, UFC 3-340-02 for blast, and the A/E/C CAD standard.
• Scale and liability. Gradon has two people. A licensed engineer of record must still seal every sheet, and A-E selection is qualifications-based under the Brooks Act, so Gradon sells to firms and does not replace them.
• Operational security of the model. A model holding many Pacific site packages becomes an aggregation target. Strict tenant isolation and no training on customer data are required, not optional.

What would have to be true

• An on-premise or air-gapped edition that runs inside a mid-tier A-E firm’s CMMC enclave.
• UFC criteria loaded as the “standards” the tool learns.
• A first contract through an AFWERX or AFCEC SBIR topic modeled on the RED HORSE airfield design award, or a DIU Commercial Solutions Opening.
• A pilot on an unclassified 1391 or minor construction package with a Pacific A-E MATOC holder before any C2 or decoy work.

Sources

Sources are linked inline throughout. Company facts come from heygradon.com, its terms and privacy pages, the YC profile and the YC launch post. Budget lines come from the FY2027 President’s Budget (PE 0208007F, PE 0603239N) and the FY2027 Pacific Deterrence Initiative book.