03 — Competitive Landscape: Who Has Tried to Make Lighting Design Accessible

"There are tools that calculate lux, and tools that make rooms look pretty. Almost nobody has built a tool that lets a Singapore homeowner feel the value of good lighting before they pay for it."

01Direct DIALux-Replacement Attempts

DIALux evo & DIALux Pro dialux.com

LIVECATEGORY LEADER ~750,000 professional users worldwide · 27 languages · Free (evo) / ~£310 yr (Pro)

What it tries to make easy

The de-facto global standard for photometric lighting calculation. Free for "evo" version; paid Pro adds workflow speed. Imports IFC from BIM and integrates with Metaroom for 3D-scanned existing spaces. Used by 450+ luminaire manufacturers to publish IES/LDT files.

What worked

Free price point + manufacturer luminaire database moat (2.5M+ fixtures). Every lighting manufacturer publishes DIALux plugins — a network effect impossible to copy. Industry-mandated for compliance reporting (SS531-style standards globally).

What failed / limits

Steep learning curve cited in user reviews; struggles with curved surfaces; high computational demands; output is engineering-grade (false-colour isolines, lux-grid tables) — appreciable only by trained eyes. No native consumer mode. Mac users locked out (Windows-only).

Why customers pick it

Because it's free and compulsory. Not because they enjoy using it.

Tech stack (best guess)

C++ / .NET · Radiosity (own engine) · IFC import · IES/LDT/EULUMDAT photometry · Windows-only · 3DS/SketchUp import

DIALux evo screenshot showing isolines and false-colour render of an interior — DIALux evo — false-colour rendering and isoline overlay. Source: dialux.com

Lesson for BSH

DIALux outputs are the credibility layer, not the customer layer. BSH's job is to wrap the credibility (lux numbers, SS531 compliance) inside something an HDB buyer can feel: a photoreal living-room render with their actual sofa.

ReluxDesktop relux.com

LIVE Free · Swiss origin · ReluxCAD for Revit ~£430/yr · Trilux/Zumtobel-aligned

What it tries to make easy

The "European alternative" — emphasises calculations, energy efficiency, and sustainability. Offers ray-tracing or real-time render output. Strong in BIM via ReluxCAD Revit plug-in.

What worked

Reviewers consistently call it "simpler than DIALux" once learned. Better Revit BIM integration than DIALux for MEP teams. Free desktop version is genuinely complete (not feature-gated).

What failed / limits

Lower brand awareness outside Europe. Smaller manufacturer luminaire library. Still pro-only — no consumer story. Same fundamental UX problem: lux numbers, not vibe.

Tech stack

C++ · Radiosity + ray-tracing · IES/LDT · Revit plug-in (ReluxCAD) · Windows

Lesson for BSH

Two free pro tools is one too many for the world. The pro market is locked. The consumer market is wide open and ungoverned. Don't compete on lux accuracy — compete on emotional clarity.

AGi32 lightinganalysts.com

LIVE USD 990/yr single user · USD 1,490/yr multi · 14-day trial · North America focus

What it tries to make easy

The North American counterpart to DIALux. Precise photometric calc + roadway/sports/warehouse specialisation. Reviewers call its renderings "valuable for communicating to clients and non-lighting professionals" — a rare admission that professional tools fail at non-pro audiences.

What worked

Standard for U.S. IESNA-compliance work. Strong roadway/exterior calc engine. Recognised in code reports.

What failed / limits

G2/Capterra reviewers complain of glitches and crashes on new versions; pricing is high vs DIALux being free. Single-user license model limits team adoption.

Tech stack

Proprietary radiosity · IES native · DWG export · CAD-style UI · Windows

Lesson for BSH

Even AGi32's marketing brags that renderings help "communicate to non-lighting professionals." Translation: the pro tool itself can't do that job. There's space for a layer that sits between pro calc and consumer perception.

ElumTools elumtools.com

LIVEREVIT-ONLY Subscription · Autodesk App Store · Lighting Analysts' Revit add-in

What it tries to make easy

The first/only fully integrated lighting calc inside Revit. Uses existing Revit geometry, materials, and family fixtures — no duplicate data entry. TM-32/TM-33 photometric format support added in 2025.

What worked

Solves the "MEP engineer redoes the model in DIALux" double-work problem. Interactive in-Revit visualisation. Marketing pitch: "great visuals for clients."

What failed / limits

RevitForum users say it lacks W/m² and uniformity outputs that Relux/DIALux have, and the result display is weaker. Locked into the Revit ecosystem — useless for non-BIM workflows. Still produces engineering-grade output for clients.

Tech stack

.NET / Revit API · Radiosity (Lighting Analysts engine, shared with AGi32) · TM-32/TM-33

Lesson for BSH

Even when the goal is "use Revit data, save time," the deliverable is still a lux grid. None of the pro tools dare to skip the lux grid. That ceiling is BSH's opportunity.

Visual Lighting + Visual Interior Tool visual-3d.com

LIVEVENDOR-OWNED Free · Owned by Acuity Brands · Browser tool exists

What it tries to make easy

Free professional photometric package + a browser-based "lumen method" calculator for rectangular rooms. The Visual Interior Tool runs entirely in-browser via HTML5 Canvas with Dropbox sync — closest the pro market gets to web-native.

What worked

Free-as-marketing model. Browser-based Interior Tool removes install friction. DWG/PDF/Visual Controls export. Native Revit plug-in.

What failed / limits

Bias toward Acuity-brand fixtures. Browser tool is "rectangular rooms only" — toy compared to DIALux evo. Still requires lumen-method literacy.

Tech stack

Desktop: Win C# · Browser tool: HTML5 Canvas + Dropbox API · IES native

Lesson for BSH

Vendor-owned free pro software trains the buyer into the vendor's catalogue. BSH already sells fixtures via bshsg.com — same playbook applies if BSH builds the tool. The luminaire becomes the funnel, not the product.

Calculux (Philips legacy) software.informer

LEGACY / DEPRECATED Last major update ~2014 · Indoor / Area / Road modules · Philips Phillum format

What it tries to make easy

Philips' in-house lighting design tool — three apps (Indoor, Area, Road) + luminaire DB manager. Once a serious DIALux competitor; now coasting as a free legacy download.

What worked

Tightly bound to Philips luminaire library. Multi-format photometric import. Free.

What failed / limits

Effectively abandoned as Signify migrated to DIALux/Relux interop. Single-vendor catalogue, dated UI. Cautionary tale: manufacturer-built design tools age out fast unless owned by an industry consortium.

Tech stack

Windows MFC · Philips Phillum + IES/EULUMDAT · Legacy installer

Lesson for BSH

If BSH builds a tool tied only to BSH stock, it dies the day BSH's catalogue narrows. The tool must accept any IES file from any vendor — even competitors. Calculux proves single-brand tools become dead weight.

ERCO Spotlight + Track Configurator erco.com

LIVECATALOGUE TOOL Free · German manufacturer · Web-based

What it tries to make easy

A web-based track-light configurator: pick line / L-shape / rectangle up to 6m, drop spots, get a wiring diagram + part list. Plus the Invia 48V configurator. Marketed via the ERCO "Handbook of Lighting Design."

What worked

Beautifully constrained scope — a single product line, a single decision tree. Output is buyable directly from ERCO. Visual web UI, not a desktop install. This is what consumer-grade lighting "design" tools should feel like.

What failed / limits

Only useful if you've already decided to buy ERCO and you already know you want track lighting. No lux output. No room context.

Tech stack

Browser JS · Probably Three.js / WebGL · Direct shopping cart integration

Lesson for BSH

ERCO's configurator is the closest thing in the pro world to what BSH needs to build for HDB owners. Constrained choices + visual feedback + direct purchase. Steal the UX pattern, replace the product with "downlights for a 3-bedroom HDB."

ChalmLite 6 (Hubbell / Chalmit) hubbell.com

LIVEHAZARDOUS-AREA NICHE Free · Hubbell-owned · Hazardous / industrial focus

What it tries to make easy

Specialist DIALux-style tool for hazardous-area lighting (oil & gas, marine). Free download. Hubbell uses it the same way Acuity uses Visual — funnel into their luminaire catalogue.

What worked / failed

Owns its niche (hazardous-area lighting designers).

Single-brand, single-niche, single-platform Windows installer. Functions as a sales tool more than a design tool.

Tech stack

Windows · IES/LDT · Hubbell catalogue · Reskinned DIALux feel

Lesson for BSH

Niche + free + manufacturer-owned is a viable model. BSH's "niche" can be "Singapore HDB & condo electrical fit-out" — narrower than DIALux's "all buildings worldwide," which makes the UX 10× simpler.

02Consumer Smart-Light Apps (Control, Not Design)

Casambi casambi.com

LIVE Free app · Finnish · Bluetooth mesh · Embedded in 300+ luminaire brands

What it tries to make easy

Tap-and-pair Bluetooth-mesh lighting control. Gallery feature lets you photograph the room and drop virtual tap-points onto each fixture in the image — closest a consumer app gets to a floor-plan UI without being one.

What worked

No hub, no gateway, no Wi-Fi dependency. Architects and specifiers love it because it's invisible (no extra hardware in the ceiling). Wins on RIBA-spec'd boutique projects. The Photo Gallery feature is genuinely consumer-friendly.

What failed / limits

Reports of setup/usability friction with certain device configs. Limited to compatible Casambi-enabled fixtures (which carry a price premium). No planning before purchase — it's a control app, not a design app.

Tech stack

BLE Mesh · iOS/Android native · Cloud backup · No hub required

Casambi app Gallery — virtual taps overlaid on room photo — Casambi's Gallery: photo of the room with tap-targets on each luminaire. Source: casambi.com

Lesson for BSH

Casambi's "photo of the room + virtual fixture taps" UI is genius and underexploited. If BSH lets HDB owners take a photo of their living room and tap where the downlights will go, they're already 80% of the way there.

Philips Hue + SpatialAware philips-hue.com

LIVECATEGORY KING ~10M households worldwide · Signify · AR floor-plan scan launched Spring 2026

What it tries to make easy

Make smart lighting feel "designed, not programmed" (their own framing). The 2026 SpatialAware update uses AR phone-scan to build a model of where each bulb physically lives, then spreads colours/brightness in patterns that mimic real light propagation. Scene gallery created by professional lighting designers.

What worked

10+ years of category dominance. Spotify integration analyses song metadata to pulse colours. Custom scenes from a photo is a killer feature for non-technical users. Pre-baked "lighting designer–approved" scenes solve the cold-start problem.

What failed / limits

Walled garden: only Hue bulbs really shine. Bridge Pro required for SpatialAware. Doesn't plan installation — only controls already-installed bulbs. Bulb-replacement model doesn't help BSH's downlight-in-ceiling buyer.

Tech stack

Zigbee 3.0 + Matter · Hue Bridge · iOS/Android · ARKit / ARCore for SpatialAware · React Native (likely)

Philips Hue SpatialAware AR scanning — Hue SpatialAware AR scan of a living room. Source: philips-hue.com

Lesson for BSH

The SpatialAware AR room-scan is exactly the on-ramp HDB owners need. BSH cannot beat Hue at bulbs, but BSH can beat Hue at built-in ceiling work — the lights Hue can't reach. Use a similar AR scan to plan downlight positions before the contractor cuts holes.

Lutron Caséta / RadioRA / HomeWorks lutron.com

LIVEPREMIUM AUDIO-VIDEO INSTALLER CHANNEL USD 200–10,000+ per home · Custom-installer-led

What it tries to make easy

Whole-home dimmer/keypad control. App lets users name and reorder rooms, organise scenes. Designed primarily to be commissioned by a pro AV installer, then handed to the homeowner.

What worked

35-year category-defining brand for residential lighting control. Premium AV installer channel guarantees a "designed" outcome. Apple Home / Google Home integration. Keypads are tactile, beautiful objects in their own right.

What failed / limits

Almost zero self-serve discovery — you find Lutron via an installer or never. The app cannot reorder devices within rooms (small but constant complaint). Premium pricing keeps it out of mass HDB market.

Tech stack

Lutron ClearConnect RF · Caseta/RA bridge · iOS/Android · Apple HomeKit / Alexa / Google

Lesson for BSH

Lutron proves that "installer commissions, homeowner controls" is a viable channel. BSH already is the installer. The missing piece is the pre-commissioning planning UI that today is delegated to engineers behind closed doors.

Nanoleaf Layout Assistant nanoleaf.me

LIVEGAMER / GEN-Z Free app · Shapes, Lines, Canvas · ~USD 200+ per panel kit

What it tries to make easy

Plan a wall mosaic before sticking panels on the wall. Built-in Layout Assistant + AR preview. Pattern library + user-generated patterns. Spotify / mic sync.

What worked

Genuine pre-purchase planning UX — rare for consumer lighting. AR preview means buyers can see panels on their actual wall before buying. Community pattern sharing creates network effects.

What failed / limits

App is "powerful but sloppily assembled" per Android Police — AR crashes, panel-count mismatches with the assistant. Aimed at gaming/streamer aesthetic, not living rooms. Limited to decorative panels, not architectural lighting.

Tech stack

Wi-Fi + Thread · ARKit / ARCore · Cloud-shared scenes · React Native

Lesson for BSH

The Nanoleaf AR layout preview is the right mechanic at the wrong price point. Adapt it: "drag downlight positions onto your AR-scanned ceiling, see lux fall on the floor in real time."

IKEA Home Smart (TRÅDFRI → DIRIGERA) ikea.com

LIVE Hub: ~USD 70 · Bulbs from ~USD 10 · Matter-bridged

What it tries to make easy

Mass-market smart lighting at IKEA prices. DIRIGERA hub replaced the older TRÅDFRI gateway in 2022. Adaptive lighting (circadian time-of-day shifts) and routines. Voice integration through Alexa / Google.

What worked

Price floor for the entire category. New app a major UX upgrade over TRÅDFRI. Matter bridge means IKEA bulbs work in Apple Home, Google Home, etc.

What failed / limits

App reviews complain about automation failures (sunrise timer breaks, lights drop offline). No design or planning surface — just on/off + colour + schedule. Pure replacement-bulb model.

Tech stack

Zigbee · DIRIGERA hub (Linux) · Matter bridge · iOS/Android

Lesson for BSH

IKEA owns the "cheapest viable smart bulb" position. BSH should not compete there. BSH's wedge is fixed installation (ceiling cuts, conduit, MCB sizing) — which IKEA explicitly will not touch.

Yeelight / Mi Home yeelight.com

LIVECHINA-LED Xiaomi ecosystem · Cheap · Two apps (Yeelight + Mi Home)

What it tries to make easy

Cheap Chinese smart lighting. Two competing apps from the same parent — Yeelight (limited, solid colours only) and Mi Home (richer, more patterns). Group control, IFTTT, Alexa/Google/SmartThings.

What worked

Aggressive pricing. Wide product range — bulbs, strips, panels, ceiling fixtures. Local Xiaomi ecosystem in Asia gives it a Singapore advantage Hue lacks.

What failed / limits

UX criticism: "app breaks UI standards throughout," "bugs lose settings when saving multiple flows." Colour matching reportedly inaccurate vs. Hue/Mi Home. Two apps confuse buyers.

Tech stack

Wi-Fi · Mi ecosystem cloud · iOS/Android · LAN control protocol (open)

Lesson for BSH

UX is the moat in this market. Premium-feeling planning + design experience beats raw feature count. A Singaporean HDB owner choosing Yeelight today is doing so on price; they'd happily pay 30% more for a "designed-for-Singapore-homes" experience.

Cync (ex-C by GE) gelighting.com

REBRANDED 2021 GE Lighting → Savant ownership 2020 · Renamed C by GE → Cync

What it tries to make easy

GE's smart-home lighting line, sold through US hardware retail. Apple HomeKit / Google Home / Alexa. Rebranded from "C by GE" after the Savant acquisition.

What worked / failed

Distribution via every Home Depot / Lowe's in America.

App is "very glitchy and buggy" per user reports. Devices "shut down randomly even when connected." Constant disconnect/reconnect cycles. C by GE → Cync rebrand was a forced response to losing brand permission from GE Appliances.

Tech stack

Wi-Fi + Bluetooth Mesh · Cync cloud · iOS/Android · Matter (newer SKUs)

Lesson for BSH

Heritage brand + retail distribution does not guarantee app quality. Buyers will switch on a $10 difference if the app frustrates them. BSH's app is allowed to be small — it must not be frustrating.

03AI / Image-Based Lighting Visualisation

RoomGPT roomgpt.io

LIVECATEGORY VIRAL HIT USD 15/mo Pro · 6M+ rooms generated · Open-source repo by Nutlope

What it tries to make easy

Upload a photo of your room → pick a style → AI redesigns the entire space in seconds. 100+ styles. Started as an open-source weekend project (github.com/Nutlope/roomGPT) and exploded.

What worked

Speed is the killer feature — no other tool matches RoomGPT's generation time. UI is minimalist. Open-source origin built credibility. App Store presence + iOS app. Before/after slider makes the value visceral.

What failed / limits

"Lacks the personalised touch of a professional designer." Best for inspiration, not buildable plans. Lighting changes are aesthetic, not photometric — no lux output. AI hallucinates furniture you can't buy. Not realistic for full remodels.

Tech stack

Next.js · Replicate API · Stable Diffusion ControlNet (img2img) · Upstash Redis · Vercel

RoomGPT before-photo of an empty living room — RoomGPT before — uploaded photo. Source: roomgpt.io

RoomGPT after — AI-redesigned living room — RoomGPT after — AI redesign in selected style. Source: roomgpt.io

Lesson for BSH

RoomGPT proves laypeople massively want "show me what my room could look like" — 6M generations. But they get fake furniture that doesn't exist. BSH can win by feeding the same vibe through real BSH-stocked fixtures, with real fittings the buyer can checkout.

Magnific AI Relight magnific.ai

LIVECREATOR / PRO Acquired by Freepik · Paid · Pro-creator audience

What it tries to make easy

Visual lighting control on any photo — place lights, adjust colour/intensity, transfer lighting style from a reference photo. Three modes: visual light placement, expert presets, reference-image style transfer.

What worked

First mainstream tool to let users physically place lights on an image via visual UI. Reference-image upload feature is differentiated. Strong product photography use case.

What failed / limits

Built for photographers and product marketers, not homeowners. No spatial geometry — it's relighting pixels, not modelling rooms. Pricing aimed at creators (~$40+/mo tier).

Tech stack

Stable Diffusion / proprietary diffusion · ComfyUI-style node graph (rumoured) · Cloud GPU

Lesson for BSH

The "place a light visually + see it change the room" UX is the entire emotional hook. BSH's tool should let homeowners drag a downlight along the ceiling and watch the floor lux change in real time — borrowing Magnific's interaction model without the photographer's hardware costs.

Adobe Firefly + Generative Fill adobe.com/firefly

LIVE USD 5/mo+ · 7 lighting style presets · Inside Photoshop and standalone

What it tries to make easy

Generative fill in Photoshop or web app. 7 built-in lighting presets when generating images. Brush-paint a region and prompt for a relight.

What worked

Trained on commercially safe imagery. Distribution through Creative Cloud's installed base. Generative fill handles lighting/perspective blending well.

What failed / limits

Community feature request: "lighting control in Firefly" is a top-voted request — users want sliders, not prompts. Today's lighting is prompt-and-pray. Still aimed at creative professionals not homeowners.

Tech stack

Firefly diffusion model · Adobe Sensei · Photoshop API · Web/Desktop

Lesson for BSH

Even Adobe's users — designers — are begging for lighting sliders, not text prompts. Prove the hypothesis: laypeople will not type "warm soft 3000K with grazing wall wash" — they want a slider labelled "cozy ←→ alert."

ChatGPT (DALL-E + GPT-4o image) openai.com

LIVE USD 20/mo Plus · Most-downloaded image tool by default of having ChatGPT

What it tries to make easy

Describe a room, get a rendering. Edit existing images with "tweak the lighting." GPT-4o now allows image editing with reference (April 2025+). Tutorials describe stacking prompts: room → lighting → décor.

What worked

Zero learning curve — write English. Infinite style variety. Tom's Guide and Coohom both publish "use ChatGPT for interior design" guides (mass-market validation).

What failed / limits

Coohom's own analysis: "ChatGPT 5.0 image generation: why your AI designs miss." Hallucinated geometry. Wall lengths off. Doors moved. Buyer can't actually purchase what's shown. Pure inspiration toy.

Tech stack

GPT-4o native image · DALL-E 3 legacy · Web + iOS/Android

Lesson for BSH

ChatGPT is the universal substitute. If BSH's tool is harder to use than "describe my living room and add warmer lights," nobody will use it. The differentiator must be: ChatGPT can't buy the fixtures. BSH can.

Decoratly decoratly.com

LIVE Web-only · 27 curated styles · Style Builder · Makeover vs Enhance modes

What it tries to make easy

Real-time before/after slider for AI room redesigns. "Enhance Mode" keeps existing elements; "Makeover Mode" rebuilds entirely. Marketed at real-estate agents and homeowners.

What worked

Slider UI is dramatically more persuasive than thumbnail grids. Real-estate vertical is the right wedge — house sellers will pay to make listings look better.

What failed / limits

No mobile app. No live-view. No furniture auto-detect. No undo/redo. Only 27 styles vs RoomGPT's 100+.

Tech stack

Stable Diffusion (likely) · ControlNet for depth · Web-only · Cloud GPU

Lesson for BSH

The real-time before/after slider is the single most persuasive UI in this category. Adopt it. Show "your living room with current lighting" vs. "your living room after BSH's plan" — same camera angle, real fixtures.

ReRoom AI + DecorMatters reroom.ai · decormatters.com

LIVEMOBILE-FIRST Freemium · ReRoom Stylefie Inc. · DecorMatters $7-99 tiers

What it tries to make easy

Photo-to-redesign on mobile. ReRoom: 20+ styles, single-photo render. DecorMatters: AR layout + community design challenges + AI design tied to actual room dimensions.

What worked

DecorMatters' angle — "real layouts, accurate proportions, believable lighting" — is closer to BSH's needs than RoomGPT's "any image goes." Community design challenges drive engagement.

What failed / limits

Mobile-only AR is shallow vs. desktop tools. AI still hallucinates buyable products. Lighting changes are cosmetic.

Tech stack

iOS/Android native · ARKit/ARCore · Diffusion model API · Furniture catalogue overlays

Lesson for BSH

DecorMatters specifically markets "believable lighting." That phrase is uncontested marketing space. BSH can claim "buildable lighting" — one notch higher.

PromeAI Relight promeai.pro

LIVE Freemium · Chinese-origin · Multi-use AI suite

What it tries to make easy

Drag light points around a photo, control brightness, get relit image. Pitched for "amateurs, architects, interior designers, product designers."

What worked

Interactive light placement (not just text prompt). Comparable to Magnific at lower price. Web-based, no install.

What failed / limits

No spatial 3D — relight is 2D pixel manipulation. No room measurements. Lower brand awareness in Western markets.

Tech stack

Diffusion + ControlNet · Web app · Cloud GPU

Lesson for BSH

The fact that Chinese SaaS is shipping interactive 2D relight tools cheap means BSH has roughly 12 months before this market commoditises. Move now or lose the timing wedge.

Marble 1.1 (World Labs) + Luma Genie worldlabs.ai · lumalabs.ai

LIVERESEARCH-GRADE Fei-Fei Li's World Labs · Luma USD 30M Series B · 3D world generation

What it tries to make easy

Marble: single image or panorama → fully explorable 3D world with improved lighting and contrast. Marble 1.1 Plus expands available 3D space. Luma Genie: text-to-3D mesh in <10 sec, includes lighting properties (metalness, roughness, emissive intensity).

What worked

Demonstrates that fully generative 3D scenes with realistic lighting are now possible. Marble explicitly calls out "interior design" as a use case. Output is explorable, not flat image.

What failed / limits

Too research-y, too expensive, too generic. Doesn't know what an HDB ceiling is. Lighting is cosmetic. No purchase tie-in.

Tech stack

Gaussian Splatting · NeRF · Diffusion · Custom training

Lesson for BSH

This tech will be commodity in 18–24 months. By 2027, an HDB owner will take an iPhone Lidar scan of their flat and a generative model will produce a Marble-style explorable 3D version. BSH should pre-build the photometric layer that sits on top — be the layer Marble can't be.

04Architectural Visualisation Tools with Lighting Modes

Enscape enscape3d.com

LIVE USD 50/mo · Owned by Chaos · Real-time, plug into Revit/SketchUp/Rhino

What it tries to make easy

One-click photorealistic real-time render directly from BIM/CAD. Real-time ray tracing on consumer GPUs. Strong interior daylit performance. VR walkthrough.

What worked

"Fast design reviews, VR/AR walkthroughs, quick client feedback" — recognised across the industry as the easy choice. Lighting and reflections rated best-in-class for interiors.

What failed / limits

Aimed at architects with Revit/SketchUp licenses — assumes upstream BIM model. Not a planning tool, a presentation tool. Output looks beautiful but isn't photometrically accurate.

Tech stack

DirectX/Vulkan · Plug-in to Revit/SketchUp/Rhino/Archicad · NVIDIA RTX-accelerated

Lesson for BSH

Enscape sets the bar for "what good interior lighting renders look like" in the architect's eye. BSH's HDB owner has never seen Enscape output. Showing them an Enscape-quality render of their own flat is itself the wow moment — it doesn't need DIALux-level accuracy to feel credible.

Twinmotion twinmotion.com

LIVE Free for <USD 1M revenue · Epic Games · Path tracing 2024+ · Drag-and-drop UI

What it tries to make easy

Unreal Engine in a drag-and-drop wrapper. Free under Epic's revenue threshold. Path Tracer added 2024 for physically correct light bounces. Four light types + emissive materials.

What worked

"Up to speed in no time" actually delivered. Free tier removes the price objection. Epic's massive 3D asset library (Megascans, Fab marketplace).

What failed / limits

"Cinematic" not "photo accurate." Path Tracer demands GPU horsepower. Still a desktop install — friction for casual users.

Tech stack

Unreal Engine 5 · Lumen + Path Tracer · Datasmith · Windows/Mac

Lesson for BSH

Real-time path-traced lighting is now free. If BSH wants the highest-quality renders, the engine cost is zero. The only investment is integrating BSH's catalogue into a Twinmotion/Unreal export pipeline.

Lumion lumion.com

LIVE USD ~1,800 perpetual / yr · Strong in exterior · ACT-3D BV (Netherlands)

What it tries to make easy

Highly photorealistic renders with cinematic effects (fog, bloom, atmospheric haze). Long the gold standard for archviz exteriors.

What worked / failed

Reviewers rate Lumion top of the trio for landscape and exterior. Strong material library.

Expensive. Steeper learning curve than Enscape. Path-traced quality, but at GPU + license cost. Interiors compete with Enscape rather than dominate.

Tech stack

Proprietary engine · Ray tracing · Windows

Lesson for BSH

Lumion's strength is exteriors, BSH's is interiors — not directly competitive. But the cinematic post-fx (bloom, glow) is exactly what an HDB owner reads as "professional photography." Steal post-process recipes, not the engine.

Unreal Engine 5 + Lumen unrealengine.com

LIVE Free up to USD 1M · Epic Games · Lumen is dynamic GI; Lyra is the shooter demo (not directly relevant)

What it tries to make easy

Engine-level access to Lumen's dynamic global illumination — no lightmap baking, just place lights. Lyra Starter Game is the customisable Unreal 5 shooter demo (sometimes mentioned in archviz tutorials but not an archviz product itself).

What worked

Lumen + path tracing produce photoreal interiors with consumer GPUs. Free tier. Massive ecosystem of plug-ins.

What failed / limits

Engine-level complexity. Need a C++ / Blueprints developer to make anything useful. Not for end-users.

Tech stack

UE5 · Lumen · Nanite · Datasmith import from CAD/BIM · C++/Blueprints

Lesson for BSH

If BSH builds a custom web-based 3D viewer (likely via Babylon.js or Three.js), the rendering recipes pioneered by Unreal Lumen are now well-documented. BSH does not need to license Unreal — adopt the techniques.

Blender + Lumen Pro / Lumiere blender.org · superhivemarket.com

LIVEFOSS Free · Cycles renderer · Lumen Pro add-on for IES photometry

What it tries to make easy

Free 3D suite. Lumen Pro add-on (developed with Wayne State interior design dept.) imports IES photometric data. Lumiere V0.4 is an open-source interactive lighting plug-in on GitHub.

What worked

Free, open-source, Mac+Linux+Windows. Cycles produces photoreal renders. Lumen Pro brings real photometry into the FOSS pipeline — extremely rare combination.

What failed / limits

Blender's learning curve is famously brutal. No consumer story whatsoever. Interior design users tend to drop out after a week.

Tech stack

Cycles + Eevee · Python add-ons · IES import (Lumen Pro)

Lesson for BSH

Blender + Lumen Pro proves the engine cost for IES-accurate consumer renders is literally zero. BSH could batch-pre-render HDB layouts overnight in Blender Cycles and serve them as static images, avoiding real-time GPU costs entirely.

SketchUp + V-Ray sketchup.com · chaos.com

LIVE SketchUp Studio bundles V-Ray · ~USD 700/yr

What it tries to make easy

The de-facto interior designer's combo. SketchUp builds geometry; V-Ray renders. Supports IES files, dome lights, HDRI maps. Post-production LightMix lets you toggle lights on/off after render.

What worked

Massive freelance interior design community. SketchUp's gentle learning curve is the polar opposite of Blender/DIALux. LightMix is genuinely clever — render once, balance after.

What failed / limits

V-Ray render times measured in minutes-to-hours per image. Not real-time. Skill cap to produce photorealism is still high. Per-image deliverable, not interactive.

Tech stack

SketchUp · V-Ray plug-in · CPU/GPU hybrid path tracing · IES native

Lesson for BSH

If BSH hires a SketchUp+V-Ray freelancer, today they can deliver beautiful one-off renders per HDB unit for ~SGD 100 each. The arbitrage opportunity is automating that into a self-serve tool — same output, zero marginal cost.

05Failed & Acquired Startups (The Goldmine)

Modsy modsy.com (defunct 2022)

SHUT DOWN JULY 2022 Raised USD 72.7M (Series C 2019) · Founded 2015 by Shanna Tellerman (ex-Google Ventures)

What it tried to make easy

Photo-real 3D renderings of your room with shoppable furniture, combined with human designer guidance ("e-design"). At its peak the most promising "online interior design" startup with $70M+ funding.

What worked

Genuine 3D photo-real renders > AI-style image generation for buyer confidence. Shoppable items closed the loop from inspiration to purchase. Built an admired designer marketplace.

What failed

Capital-intensive: each rendering required human-in-the-loop modelling. CEO cited "capital restraints and a shaky market" + a failed acquisition. Customers were left with unfinished projects and awaiting refunds — a brand-disaster ending. Designers received notice only one week before contracts cancelled. Couldn't sustain unit economics; could not scale automation fast enough to outrun cash burn.

Tech stack

Web app · 3D-modelled product catalogue · Human designer marketplace · Likely Three.js + custom render farm

Lesson for BSH (the biggest one in this report)

Modsy proved demand for "show me my real room rendered with real products" but died on unit economics. BSH already has the human-in-the-loop covered (Wei Xiong / Abby on engineering, BSH's own installers). The render layer needs to be automated and cheap; the human curation already exists at no incremental cost. This inverts Modsy's failure: don't build a tool that needs designers — build a tool that helps existing designers (BSH staff) move 10× faster.

LIFX lifx.com (acq. 2022)

WENT BUST APRIL 2022 · ASSETS ACQ. BY FEIT AUG 2022 Australian · Pre-Hue colour-changing smart bulb pioneer

What it tried to make easy

Wi-Fi colour smart bulbs without a hub (a clear differentiator from Hue's bridge requirement). HomeKit, Alexa, Google. Strong brand among enthusiasts.

What worked

Hub-free Wi-Fi was a real differentiator. Bright colours. Loyal community.

What failed

Hue's distribution and Bridge ecosystem won. LIFX bulbs cost more than equivalent Hue. Wi-Fi mesh saturation problems with 20+ bulbs. Lifi Labs Inc. ran out of cash April 2022. Feit Electric (US private-label retailer) bought the assets to keep the brand and software alive but stripped most engineering.

Tech stack

Wi-Fi (no hub) · iOS/Android · Cloud + LAN protocol · Custom firmware

Lesson for BSH

Even a great product dies if it can't out-distribute Hue/Signify. BSH should not try to be a smart-bulb brand. BSH's distribution is its own contractor network and bshsg.com — and that's enough to win Singapore HDB without ever needing global retail.

Noon Home noonhome.com (acq. 2020)

ACQ. BY RACEPOINT / SAVANT JAN 2020 Raised ~USD 50M Series A-B · Founded 2017 · Silicon Valley · The "Nest for light switches"

What it tried to make easy

Replace ugly dimmer switches with a beautiful capacitive-touch screen that learns scenes. Reviewed by TechHive as "the very best smart switch for your home, and it's priced accordingly." Premium price ($150–200/switch).

What worked

Beautiful hardware. Pre-baked "good lighting scenes" without needing to think. Won design awards. Got a "Nest of light switches" narrative going.

What failed

$150 per switch × 10 switches per home = $1,500 just for switches. Couldn't crack mass market. Software was tied to Noon's cloud, fragile. Sold to Racepoint Energy (sister of Savant) for an undisclosed amount — likely a soft landing rather than a win.

Tech stack

Custom touchscreen hardware · Cloud + LAN · iOS/Android · Apple HomeKit

Lesson for BSH

Premium hardware at the switch is a $50M cautionary tale. Don't build hardware. Build the planning layer. Let the customer install whatever switches they want (or BSH's own off-the-shelf ones).

Finally Light Bulb / Lucidity Lights finallybulbs.com (defunct)

DEFUNCT ~2019 Raised USD 23M+ · Boston · "Acandescent" patented induction tech

What it tried to make easy

A "real incandescent glow" LED replacement using patented induction technology. Marketed as "the bulb that looks exactly like the one Edison made, but 75% more efficient." Crossed Crowdfunder darling → retail.

What failed

LED CRI and warm-white quality caught up to Finally's induction tech faster than Finally could scale manufacturing. Premium pricing (~$10/bulb) collapsed when Hue/LIFX/Cree drove down quality smart-bulb prices. "Better incandescent glow" turned out not to be a feature buyers paid for. Quiet wind-down.

Tech stack

Patented induction light source · No app · Edison-screw form factor

Lesson for BSH

Even patented IP doesn't save you if commodity LED reaches "good enough" before you reach scale. BSH should not bet on proprietary tech — bet on the buying experience around commodity LEDs.

Soraa soraa.com (acq. 2020)

ACQ. BY ECOSENSE 2020 Raised USD 175M+ · Shuji Nakamura (Nobel) co-founded · Gallium nitride on gallium nitride LEDs

What it tried to make easy

"Skip the smart home and focus on the science of the color spectrum." Bulbs with the highest CRI on the market (~95+) and a UV "fluorescence enhancement" that made whites whiter. Premium positioning to museums, hospitality, retail.

What failed

Brilliant tech, no consumer story. CRI 95+ at $40/bulb is invisible to a homeowner who can't tell CRI from CCT. Pivoted to commercial-only, then sold to Ecosense. A textbook "great tech, no narrative" outcome — the buyer never felt the value of CRI 95+.

Tech stack

Proprietary GaN-on-GaN LED · Full visible spectrum · Premium MR16/PAR fixtures

Lesson for BSH (huge one)

Soraa had the best lighting in the world and still failed because they could not make buyers feel the difference. BSH's hypothesis — "make DIALux output appreciable to laypeople" — is literally the unsolved problem Soraa died trying to solve. Don't sell lux; sell the perceived difference between "ugly HDB lighting" and "magazine-cover HDB lighting."

LBNL Radiance + pyradiance + frads github.com/LBNL-ETA

LIVE — Lawrence Berkeley National Lab Free FOSS · Validated lighting simulation tool · 30+ years of research

What it tries to make easy

The validated scientific reference for lighting simulation. Used by researchers and high-end consultants. pyradiance wraps it in Python. frads couples Radiance with EnergyPlus for annual daylight simulation.

What worked

Scientifically validated against measured data — courts and standards bodies accept Radiance outputs. Free, FOSS, Python-scriptable.

What failed / limits

UX is "1990s command-line Linux." Renders take hours. Effectively zero consumer adoption. Lives in PhDs' laptops.

Tech stack

C · Command-line · Python wrappers (pyradiance) · Honeybee/Ladybug for Grasshopper

Lesson for BSH

If BSH ever needs provably accurate lux numbers for a Singapore SS531 compliance argument, pyradiance is the free engine that backs it. Use it as a backend "ground truth" check on whatever flashy frontend BSH builds.

06Furniture / Interior-Design Sites (Lighting as Side-Quest)

Houzz Pro Room Planner houzz.com

LIVE 25M+ photos · 100K+ ideabooks · Designer marketplace

What it tries to make easy

The all-in-one workspace for interior designers: moodboards, 3D floor plans, clipped products with pricing/dimensions, AR walkthroughs. Lighting tips published as ideabooks but no calc tool.

What worked

SEO dominance on "kitchen lighting ideas" and similar searches. Designer-led network effects. AR walkthrough mode is genuinely useful for late-stage approvals.

What failed / limits

Lighting treated as a decorative product category, not a physics topic. No lux, no glare, no compliance. "Pro" tier is the actual planner — consumer side is browse-and-pin.

Tech stack

Web app · 3D floor plan engine · iOS/Android · Product catalogue · React

Lesson for BSH

Houzz teaches the buyer that "lighting" means "fixtures and lampshades" — not "lux and beam angle." BSH's tool needs to bridge these two worlds: speak the buyer's language (vibe, brightness, mood) but deliver the contractor's plan (positions, wattages, MCB count).

Wayfair View in Room 3D wayfair.com

LIVE In-app · ARKit/ARCore + LiDAR · Realtime virtual lighting and shadows

What it tries to make easy

"AR with a purpose." Tap a product in the Wayfair app → place it in your room at scale → it casts real shadows based on your room's actual lighting via LiDAR + RealityKit.

What worked

Realtime accurate shadows based on actual room lighting is a genuine quality leap. Drives Wayfair conversion. Hides the AI/ML complexity behind a "Tap here to see it in your room" button.

What failed / limits

Furniture, not architectural lighting — a floor lamp, not a downlight. iOS-LiDAR-only for the good version. Doesn't simulate replacing the room's own lighting.

Tech stack

iOS RealityKit + LiDAR · Android ARCore · USDZ/glTF assets · Cloud asset CDN

Lesson for BSH

If a sofa app can render real shadows in AR, a downlight app can render real lux in AR. The technology gap is zero. The only missing piece is photometric data + an honest installer telling buyers what's actually buildable.

IKEA Place AR ikea.com/place

LIVE (since 2017) Free · 3,200+ IKEA products · Apple ARKit launch app

What it tries to make easy

One of the first mass-market AR shopping apps. Place an IKEA item in your room at "98% accuracy" in scale, fabric, lighting, and shadow.

What worked

Massive download success — became the #1 free non-game AR app at launch. IKEA chatbot for navigation. "Room Sets" feature places multiple items at once.

What failed / limits

"Struggles with scale" — items balloon from tiny to gigantic when moved around. Old units don't have LiDAR so it relies on ARKit floor detection. Doesn't model lighting changes.

Tech stack

ARKit / ARCore · Unity (likely) · 3D-modelled IKEA catalogue

Lesson for BSH

IKEA Place trained an entire generation to expect AR-in-room shopping. The expectation is now table stakes. BSH's tool that doesn't have AR will feel old-fashioned, even if the underlying physics is far better.

West Elm Design Crew Room Planner westelm.com

LIVE Free with account · Pottery Barn / West Elm / WS Home / PB Teen cross-brand

What it tries to make easy

"The Sims for designing room layouts." Measure room → drag-drop West Elm furniture (or generic) → save layouts → buy.

What worked

Free with account, no install. Apartment Therapy review: "saves so much time." Cross-brand catalogue. Direct conversion to checkout.

What failed / limits

Lighting handled as a furniture category (table lamps, floor lamps). Zero photometric simulation. Pretty 2D/3D plan, not a render.

Tech stack

Three.js (likely) · Web · Product catalogue API · Save-to-account

Lesson for BSH

The "measure → drag → buy" loop is now a commodity UX. BSH's differentiator is that installation follows the buy click. West Elm needs you to schedule a moving truck; BSH books an electrician.

Planner 5D + Coohom + HomeByMe planner5d.com · coohom.com · homebyme.com

LIVE Freemium · Mass-consumer DIY · Coohom claims 1M+ 3D models

What it tries to make easy

The triumvirate of consumer-grade 3D room planners. Drag-drop walls, doors, furniture; toggle 2D/3D; render. Coohom: photo-real 4K render in <10 sec; 720° walkthrough mode. Planner 5D: "Bernard" AI design assistant + Smart Wizard for instant auto-furnishing.

What worked

Mass appeal. Coohom's render speed is industry-leading. AI-assisted layout fills in the cold-start problem. Mobile-first.

What failed / limits

Reddit complaints: lag on desktop, paywalls for premium content, no snapping for precision. Lighting available as a fixture but not photometrically simulated. Outputs are pretty, not buildable.

Tech stack

Three.js / Babylon.js (suspected) · Cloud render farm · Mobile + web · AI-assisted layout (Planner 5D Bernard)

Lesson for BSH

The mass consumer is already comfortable dragging walls and dropping fixtures in a 3D room. BSH does not need to teach this gesture — it's table stakes. But none of these tools answer "if I put 6 downlights at 6W each in this 12m² living room, will it be too bright?" That's the unfilled gap.

Sweet Home 3D + Live Home 3D sweethome3d.com · livehome3d.com

LIVE Sweet Home 3D free FOSS Java · Live Home 3D paid Apple ecosystem

What it tries to make easy

Sweet Home 3D: 20+ years of FOSS interior design app, Java-based, beloved on Hacker News. Live Home 3D: native Apple/Vision Pro variant with daylight + artificial lighting, geographic-location-driven sun simulation.

What worked

Free (Sweet Home). Native Mac/iOS/Vision Pro polish (Live Home). Both simulate time-of-day daylight which most competitors skip.

What failed / limits

G2 reviewers call Sweet Home "a bit clunky, not for professional design." Both lack lux calc — lighting is purely visual. Live Home 3D is locked into Apple devices.

Tech stack

Sweet Home 3D: Java + Java3D · Live Home 3D: Swift + Metal · Both desktop-first

Lesson for BSH

Even free 20-year-old FOSS room planners simulate daylight. There's no excuse for BSH's tool not to model Singapore's actual sun path (~1°N latitude) — and that small detail will make every render feel local in a way Western competitors can't fake.

Havenly havenly.com

LIVE USD 149–599 per room · Marketplace of human designers · Acquired The Inside (2023), Interior Define (2023)

What it tries to make easy

Pay a fee → get matched with a human interior designer → receive a moodboard + shoppable product list. Survived Modsy's death by being human-led, not 3D-render-led. Picked up Modsy's customers when it collapsed.

What worked

Lower CAPEX than Modsy's 3D pipeline — humans use Photoshop moodboards, not photoreal renders. Acquired Interior Define and The Inside in 2023, pivoting toward DTC furniture vertical integration. Still standing while Modsy is gone.

What failed / limits

No real 3D / lighting simulation — just moodboards. Hard to scale beyond the human designer count. Customer churn after the room is "done."

Tech stack

Web app · Designer marketplace · Shopify-style commerce · Probably Rails / React

Lesson for BSH

Havenly survived because they didn't try to automate the designer out of the loop. BSH already employs the designer (Wei Xiong, BSH engineers). The tool's job is to amplify them, not replace them. Modsy died trying to replace; Havenly lives by amplifying.

07Comparison Matrix · All Products

A single table view. Rows are sorted roughly from highest accuracy / lowest accessibility (top) to highest accessibility / lowest accuracy (bottom). The "sweet spot" BSH is aiming at — high accessibility and photometric credibility — is mostly empty in this table.

Product	Target audience	Tech approach	Status	Reach	Lesson for BSH
DIALux evo	Pro lighting designers, MEP engineers	Desktop radiosity + IES library	Live · category leader	~750K users	Wrap its credibility, hide its UX
Relux	Pro EU designers, Revit MEP teams	Desktop radiosity + ReluxCAD plug-in	Live	Smaller than DIALux	Don't compete on accuracy
AGi32	US lighting designers	Desktop, ~USD 990/yr	Live	US-heavy	Even AGi32 admits "renders communicate to non-pros" — gap exists
ElumTools	Revit MEP shops	Revit add-in radiosity	Live · niche	BIM market	Even in BIM the deliverable is still lux grid
Visual Lighting	Pro + browser hobbyist	Desktop + HTML5 Canvas browser tool	Live	Acuity ecosystem	Vendor-owned free pro tool is a viable funnel
Calculux	Philips spec'd projects	Legacy Windows	Deprecated	Shrinking	Single-vendor tools age out
ERCO Configurator	Track-lighting buyers	WebGL · constrained UX	Live	ERCO buyers	Closest pro-world UX to consumer needs
ChalmLite	Hazardous-area lighting	Windows · Hubbell catalogue	Live · niche	Hazardous market	Manufacturer-niche works
LBNL Radiance	Researchers, PhDs	FOSS C · Python wrappers	Live · scientific	Tiny	Free engine for ground-truth back-end
Enscape	Architects on Revit/SketchUp	Real-time ray tracing plug-in	Live	Huge in archviz	Sets quality bar for what "good" looks like
Twinmotion	Architects, designers	UE5 Lumen + Path Tracer	Live · free tier	Massive	Free photoreal engine
Lumion	High-end archviz	Proprietary engine · USD 1,800	Live	Premium	Borrow cinematic post-fx
Unreal + Lumen	3D devs	Engine · C++/Blueprints	Live · free	Universal	Engine cost is now zero
Blender + Lumen Pro	FOSS designers	Cycles + IES add-on	Live · free	Niche	Free Cycles batch-render pipeline
SketchUp + V-Ray	Freelance interior designers	CAD + render plug-in	Live	Large freelance pool	SGD 100/render arbitrage
Magnific Relight	Pro photographers, marketers	Diffusion + visual UI	Live	Creator class	Steal the visual light-place UX
Adobe Firefly	Creative pros	Diffusion + generative fill	Live	20M+ Creative Cloud	Users beg for sliders not prompts
PromeAI	Mixed pro/amateur	Diffusion + light-place	Live	Growing	Chinese SaaS clock is ticking
Casambi	Architects via fixture spec	BLE mesh + photo overlay app	Live	300+ brands	Photo-overlay UI is gold
Philips Hue + SpatialAware	Consumer enthusiast	Zigbee + AR scan	Live	~10M households	AR scan is the consumer on-ramp
Lutron Caséta	Premium AV installer channel	RF + keypad	Live	US affluent	Installer-led model works
Nanoleaf	Gen-Z, gamers	Wi-Fi panels + AR layout	Live	Streamer audience	Pre-purchase AR planning is right
IKEA Home Smart	Mass market budget	Zigbee + DIRIGERA hub	Live	Massive	Don't compete on bulb price
Yeelight / Mi Home	Asia budget consumer	Wi-Fi + Xiaomi cloud	Live	Big in Asia	UX moat in this market
Cync	US hardware retail	Wi-Fi + cloud	Rebrand 2021	Mass retail	Heritage + retail ≠ app quality
RoomGPT	Homeowners, casual	SD ControlNet img2img · Vercel	Live · 6M+ rooms	Huge demand signal	Same vibe, real BSH fixtures
Decoratly	Real-estate + homeowners	Web diffusion + slider	Live	Growing	Adopt before/after slider
ChatGPT image	Anyone	GPT-4o native image	Live	Universal	The substitute baseline
ReRoom / DecorMatters	Mobile-first consumers	Diffusion + AR	Live	Mass-mobile	"Believable lighting" is open marketing space
Marble 1.1 / Luma Genie	Researchers, devs	Generative 3D · Gaussian splatting	Live	Early adopters	Be the photometric layer Marble can't be
Modsy	Affluent US homeowners	3D model + human designer	Dead 2022	Was sizeable	Automate the render, keep BSH's humans
LIFX	Smart home enthusiasts	Wi-Fi bulbs no hub	Acq. by Feit	Was global	Don't be a bulb brand
Noon Home	Premium US homeowner	Touchscreen switch + cloud	Acq. by Savant	Premium niche	Don't build switch hardware
Finally Bulb	Nostalgia / quality buyer	Induction LED	Defunct ~2019	Small	Don't bet on proprietary tech
Soraa	Premium / hospitality	GaN-on-GaN LED · CRI 95	Acq. 2020	Premium niche	Selling lux fails; sell perceived change
Houzz Pro	Interior designers	Web + 3D plan + AR	Live	Huge SEO	Bridge their language to contractor's
Wayfair View in Room	Furniture shoppers	ARKit + LiDAR + RealityKit	Live	Massive	Lux-in-AR is now technically trivial
IKEA Place	IKEA shoppers	ARKit · USDZ assets	Live since 2017	Massive	AR is table stakes now
West Elm Design Crew	Furniture shoppers	Web 3D + catalogue API	Live	Mass	Measure→drag→buy is commodity
Planner 5D / Coohom / HomeByMe	DIY consumers	Three.js · Cloud render · AI assist	Live	Mass	Add the "is it too bright?" answer
Sweet Home 3D / Live Home 3D	Hobbyists	Java / Swift desktop	Live	Long-tail	Simulate Singapore's actual sun path
Havenly	Online design buyers	Designer marketplace + DTC furniture	Live	Niche	Amplify designers, don't replace them

Where the White Space Is — What Nobody Has Done Well That BSH Could

Across 38 products, six recurring gaps emerged. Each is a defensible position. Together, they form the BSH thesis.

Gap 01 · Photometric × Photoreal

Lux that is also a render

DIALux / Relux / AGi32 give true lux but ugly false-colour output. Enscape / Twinmotion / RoomGPT give beautiful renders but no photometry. Nobody at consumer accessibility shows both in one frame: "Here's your living room at SS531-compliant 100 lux average — and here's what it'll look like in real life." That overlay is the BSH product.

Gap 02 · Geo-localised

Singapore sun path, HDB ceiling heights, Singapore-stock fixtures

Every tool assumes generic geometry. None ship with HDB BTO floor-plan templates, 2.6m ceiling presets, Singapore latitude sun path, or the fixtures BSH actually stocks. The Western 3-bedroom planning UI does not map to a Singapore 4-room flat. This is a 100m moat once built.

Gap 03 · Buy-the-render

The fixtures in the render are the ones you check out with

RoomGPT shows fake furniture. ChatGPT hallucinates. Modsy renders had real items but died on unit economics. None of these tools tie the visual back to a one-click order of installable fittings. BSH already runs bshsg.com — the missing layer is a render that flips into a basket. This is unique to BSH having both the catalogue and the installation arm.

Gap 04 · "Vibe" → "build"

Slider in, drawing out

The consumer says "cozy" or "alert." The contractor needs "8 × 6W 3000K downlights at 1.2m grid, two MCB 16A circuits." No tool today does that translation. ChatGPT can fake the consumer side; DIALux can fake the contractor side. Bridging them is a pure-software problem with an asymmetric payoff — the consumer doesn't see the lux number; the engineer doesn't see the vibe slider.

Gap 05 · Pre-purchase AR planning

Drag downlights on your ceiling, see the floor light up

Hue SpatialAware does AR after the bulbs are installed. Nanoleaf does pre-purchase AR but for decorative panels. Wayfair does AR but for sofas. Nobody does pre-installation AR for built-in ceiling lights — which is exactly what HDB owners face during renovation. The phone scan + photometric overlay is technically trivial in 2026; nobody has bothered because the audience felt small. It isn't.

Gap 06 · Trust by track record

"BSH did 12,000 HDB units. Here's what worked."

Every tool starts from a blank slate. BSH has 35 years of installations and bshsg.com. A tool that says "Here's how 4-room Edgefield units typically light up; here's our recommended layout from 2,000 such jobs; click to customise" leverages an asset no startup has. This is the harness layer (per Anthropic's principle) — independent verification that no fresh entrant can fake.

The minimum viable wedge: A web-based tool a Singapore HDB owner uses for 5 minutes — uploads a phone photo or picks their BTO floor plan, drags one slider labelled "Cozy ↔ Bright," sees a photoreal living-room render update in real time, and gets a printable plan of fixture positions + a "Buy & Schedule Installation" button that lands in bshsg.com.

That product does not exist anywhere in this 38-entry landscape. The technology to build it (Three.js + Stable Diffusion + IES files + Singapore sun path + bshsg.com catalogue API) all exists for free or cheap. The defensibility is BSH's installer network — which no software competitor can replicate without years of physical operations.