Simam AI Lab is the research division of Simam Digital. We take new AI research, from depth and 3D to vision and documents, and turn it into working demos you can open, with measured results and honest caveats.
Every result is scored against something the model never saw. Drag the images, probe the chart, then read how each one was done.
measured in 3D from 16 seconds of drone footage, with a sample register of who has every flat, shop and office, what the empty plots could hold, parking, security, rooftop solar and shadows
Your plaza or office block measured in 3D, your own register of sales, bookings and tenants on the floors it belongs to, and parking, security and solar planned on the real site.
A developer selling a commercial block is asked the same questions every day: who has which floor, what can still be built, where will people park, is it safe, and what will the electricity cost. We took 16 seconds of stock drone footage over the entrance of a housing society in Pakistan, rebuilt it in 3D and measured every building from the 3D itself: footprint, height and floors. On top sits a sample register, every flat, shop and office marked sold with possession, on instalments, booked with token money, rented or still for sale, and the empty plots with what a by-law would let a developer build. Parking is set against what the buildings need, including the Friday rush at the mosque; sample cameras show what they can see; every roof gets its solar yield; and a sun study casts the buildings' shadows through the day, with north estimated from the mosque, which faces the qibla. From the drone's own position, a divider slides between the 3D and AI pictures of the empty plots a year on and built out.
Caveat. The register (owners, files, instalments, tenants and rents), the by-laws, the parking standards, the cameras and the solar figures are sample data and rough estimates, labelled on every view. What each building is and its floors are read by eye. The 3D was trained on all 33 frames, so no frames were held back to score it. The society is not named, and its name was taken out of the 3D.
Each demo runs in real time on the visitor’s own device. No install, no account.
The entrance of a housing society in Pakistan in 3D from 16 seconds of drone footage: every building measured, who has every flat, shop and office, what the empty plots could hold, parking, security, rooftop solar and shadows.
A housing society on its first day, bare graded land and a site office, in 3D from one drone orbit. An AI masterplan is laid on the real ground, with every street, pipe, cable and street light traced from it, and a timeline that builds the society phase by phase.
A housing society in Pakistan in 3D from one drone clip: every plot with its build stage, owner, file and payments, roads and a services plan underneath, and a timeline that raises the society building by building.
Ten handwritten Met Office rainfall sheets, one from each decade since the 1860s, read by four open OCR models and checked cell by cell against volunteers' copies. Hover any cell to see who read what.
The Giza plateau in 3D from one drone clip. Press a year and the pyramids rise in the order they were built, beside Gemini's pictures of the same view, while hologram crews haul blocks along the plateau's real routes.
One 44-second drone pass over a housing site, rebuilt in 3D, with every plot the detector found pinned where it stands and coloured by stage. Click a pin to see the frame it was seen best in.
Ordinary video lifted into a 3D memory you can walk around: people and machines as light, measured in metres from the people themselves. A forklift near miss, an anonymised concourse, miners in 1901, and a small vision model that narrates and bets on the next two seconds, scored against what really happened.
Upload a clip and every moment becomes a slice with AI depth, stacked into a block of time or strung along the path the camera flew, measured from the video itself. Type "blue car" and its path is drawn through time, while a small vision-language model says what the scene is. Nothing is uploaded.
A flat video of our factory cell rebuilt in space and time. Scrub the clip, freeze a moment and walk round it, or watch every moving point build up in the order it was captured, with a capture report that measures each frame and finds the moments worth jumping to.
A flat 360° clip turned into stereo 360° for headsets: an AI model estimates distance in every direction and we build both eyes for the whole sphere. Drag to look around in red/cyan, or open it on a Quest.
A Gaussian-splat scene cut into a quality ladder and streamed like video: a first view in half a second, sharper versions as the connection allows. Throttle the connection and watch it adapt.
An AI depth model runs on your own graphics card and a shader builds a left and right eye from every frame. Red/cyan, side by side for headsets, or wiggle 3D. Try the sample clip, your webcam or your own video.
A 3D avatar built from one photo copies your movements through your webcam. Pose estimation runs inside the page, and our retargeting turns 33 body landmarks into bone rotations, live.
Eight generated humanoid robots and workers carrying out tasks in a real-time factory cell. Characters, props and motion were produced by an AI pipeline, then auto-rigged and retargeted.
A reproduction of RT-Splatting on a published vehicle scene, separating glass reflections from what is seen through the glass. Compare each rendered pass against the held-out photographs.
A single car photograph becomes a textured 3D vehicle with its wheels and lamps separated automatically, then drives in the browser. Petrol or electric, automatic or manual, day or night.
No model inside yet: these simulate the task and score how it was done, so an assessor or an AI coach has something real to judge.
A gantry crane whose load hangs on a physics joint, so it swings like a real pendulum. Stop the traverse on the pendulum's period and the load is left swinging 1°; stop half a period out and it is 18°, with the same peak either way.
A lock-out/tag-out trainer on a pressurised line. Valves take held effort and spring back if released, the line only bleeds down once both ends are shut, and unsafe acts are recorded rather than blocked: finish every step and still fail.
The wider studio work our research feeds into: spatial platforms, dashboards and free business tools.
The lab is not presented as a generic AI catalogue. Every project is organised around a repeatable loop that connects spatial evidence, domain models, human interfaces and deployment constraints.
In development / Simam LM · AI harness design · engineering assistants
Physical environments become queryable evidence.
Models are selected around the work, not the other way around.
Autonomous steps remain inspectable, reversible and accountable.
Systems can run close to sensitive operational data.
Reusable industry agent systems for construction, museums, engineering, infrastructure and public services.
Research into how AI understands Gaussian Splats, BIM, GIS, Digital Twins and the physical world.
Multi-step systems where drones, vision models, agents and human approval produce operational outputs.
Private, edge and on-premise AI for organisations that need control over sensitive knowledge and data.
Voice, gesture, XR, digital twins and spatial interfaces for working with AI inside real environments.
Poster-first reels keep the archive fast. Full video loads only when a viewer asks for it.
An independent R&D concept exploring how a conversational, geospatial twin could connect healthcare estate, flow, scenarios and room-level context.
A browser-based spatial viewer for exploring stadiums, surrounding places and the stories attached to them.
An infrastructure command interface combining live mapping, site intelligence, operational context and AI-assisted decisions.
Testing how a headset-side assistant can interpret scene context and return short operational guidance.
A spatial agent searches captured site geometry for hazards, barriers and inspection targets.
Comparing local LLM responses against engineering notes without moving sensitive files into a shared cloud tenant.
Drone imagery flows through vision analysis, risk extraction, report drafting and human approval.
Selected prototypes and live products that show the lab direction in practice: spatial maps, digital twins, AI assistants and operational command layers.
An independent Simam Digital R&D concept exploring how a conversational, geospatial operations twin could connect estate, flow, scenarios and room-level simulation. It is not an official NHS app or NHS-endorsed service.
Problem
Healthcare operations are spread across maps, dashboards, spreadsheets and site-level tools, making it difficult to carry context from system-wide pressure to a local operational decision.
Prototype
A public proof of concept with a national 3D estate view, simulated operational indicators, scenario planning, decision-support research and an assurance surface that explains the direction of a real integration.
What we learned
The map earns attention, but the useful research lies in connecting place, pressure, assumptions, evidence and human review in one legible workflow.
Next research question
How should the twin be evaluated with qualified operations and clinical-safety stakeholders before any real-data pilot is considered?
A photorealistic 3D stadium field guide that turns venues into explorable, story-led spatial products.
Problem
Flat venue maps and static photography struggle to communicate scale, access, atmosphere and location context.
Prototype
A live browser-based stadium viewer with photoreal 3D, guided orbit interactions, venue facts and curated story layers.
What we learned
People understand a venue faster when navigation, facts and stories sit directly inside the spatial interface.
Next research question
Can an AI guide personalise tours for accessibility, fan routes, commercial hospitality and operations teams?
A smart corridor concept for planning, monitoring, risk alerts and digital twin administration across complex highway projects.
Problem
Major infrastructure programmes split schedules, risk registers, site evidence and reporting across disconnected tools.
Prototype
A corridor command interface combining 3D mapping, progress state, drone feed evidence, hazard alerts and project controls.
What we learned
Operators need fewer dashboards and more decision context: what changed, why it matters and what action is due next.
Next research question
Which workflow should the AI own first: weekly reporting, risk triage, inspection planning or client summaries?
A 3D live-map interface for venues, emergency access, routing, asset intelligence and AI-assisted operational reporting.
Problem
Large venues and event sites need shared situational awareness across access, safety, crowd movement and asset state.
Prototype
A concept command layer with Google Maps-style 3D context, emergency routes, assistant prompts and operational tabs.
What we learned
The strongest interface pattern is not a chatbot beside a map; it is an assistant that can act on spatial objects.
Next research question
How should spatial agents verify answers when the source is a live map, a Gaussian Splat, a drone feed or a human report?
A construction intelligence interface for project visibility, safety actions, phase planning, alerts and exportable reporting.
Problem
Construction teams need to coordinate projects, assets, safety tasks and reports without losing the field reality behind the paperwork.
Prototype
An internal command-center demo linking active projects, map assets, safety alerts, phase planning and report export actions.
What we learned
A credible construction copilot needs human approval points, traceable evidence and role-specific views before automation.
Next research question
What is the smallest useful agent team for a live project: planner, safety officer, cost manager or reporting assistant?
A free browser workspace that helps SME leaders move from AI uncertainty to a costed, structured transformation plan.
Problem
Teams are often asked to invest in AI before they can explain their readiness, priority use cases, expected return or the system needed to deliver it.
Prototype
An integrated hub with readiness context, AI Pulse, architecture whiteboarding, app design, ROI modelling, roadmaps and persona-aware guidance.
What we learned
The valuable unit is a chain of decisions rather than a single answer: assess, model, design, sequence and then decide where expert help is needed.
Next research question
Can a strategy output become a reusable Simam blueprint without losing its assumptions, evidence and human decisions?
A no-signup suite of free tools for understanding AI readiness, modelling ROI and finding a proportionate first step.
Problem
Smaller organisations know AI matters but often cannot see whether their immediate constraint is leadership, data, systems, skills or a defined use case.
Prototype
Client-side tools covering a 20-question readiness assessment, ROI calculator, use-case finder, jargon buster and AI-type recommendations.
What we learned
Useful guidance earns more trust when the first interaction is genuinely valuable, private by default and does not require an email address before showing a result.
Next research question
Which readiness signals best predict a successful first project, and how should the tool recommend the next step without pretending to replace discovery?
Two practical tools from Simam Digital for organisations deciding where AI and digital transformation can create useful, measurable change. They are free to explore and designed to make the first conversation more grounded.
A free strategy and digital-transformation workspace for exploring AI trends, system architectures, ROI assumptions and practical routes from idea to implementation.
A five-minute, no-sign-up assessment that helps a business understand where it stands before investing in AI, then turns the result into a prioritised action plan.
An independent healthcare operations concept connecting estate context, simulated pressure, scenario planning and room-level digital twin research.
Open research recordA public depth-to-3D workbench that exposes provenance, uncertainty, portable outputs and reproducibility evidence alongside the visual result.
Open research recordA photoreal venue exploration platform showing how spatial interfaces can carry place, story, wayfinding and future AI-guided interaction.
Open research recordEleven chapters on why we exist, what we build, and where applied AI, spatial computing and Digital Twins take the physical world by 2035. Written for clients, partners, funders and future colleagues.
Simam AI Lab is the research and innovation division of Simam Digital Ltd. The lab turns working prototypes, spatial interfaces, AI workflows and digital twin experiments into practical evidence for organisations managing complex physical environments.
Founder-led by Junaid Malik, a product engineer and technical lead with more than 11 years of hands-on delivery across product strategy, architecture, interaction design and implementation.
Visit Simam DigitalCompany
Simam Digital Ltd
Location
Wakefield, UK
Focus
Applied Spatial Intelligence
Channels
GitHub + LinkedIn
Serious enquiry
University, council and enterprise research around AI agents, digital twins, local intelligence and spatial workflows.
Serious enquiry
Serious walkthroughs for teams evaluating spatial AI, operational copilots, private AI or prototype-to-product work.
Serious enquiry
Applied pilots for infrastructure, cultural heritage, education, public services, transport and regional innovation programmes.
Prefer a structured brief?
Use the proposal flow when you already know the research area, timeline, budget range or organisation type.
We are actively seeking university, combined authority and industry collaborators for applied AI, Digital Twin and XR pilots.