Skip to content

CapabilitiesFive practices

What we actually do

Not a service menu. Five practices that keep showing up in the same projects, listed with the parts we hand over and the parts we refuse to fake.

01AR, VR, and spatial computing

Immersive

The original 615io practice: AR and VR built for training, medicine, and product education, where scale and presence teach something a slide deck can't.

We have put thirteen-foot mining tires in a living room and a child's heart in a surgeon's hands. Immersive work earns its budget when physical scale, spatial understanding, or presence is the actual lesson. Everything else — the capture pipeline, the optimisation passes, the device rollout — exists to protect that one moment of understanding.

Where we say noWe will not build a headset app because a competitor has one. If the lesson survives a video, we will tell you to shoot the video.

Deliverables

  • WebXR, WebAR, and native headset builds
  • 3D pipelines: scan, model, optimize
  • Real-time engine development
  • Volumetric and photogrammetry capture
  • Training and simulation design
  • Device strategy and rollout support

StackUnity · Unreal · WebXR · three.js · RealityKit

02Software that does the work

Agentic Systems

LLM-powered agents wired into the tools your team already uses — scoped, evaluated, and shipped to production rather than left in a demo notebook.

We design agents around a real job to be done, not around a model. That means explicit tool boundaries, human checkpoints where the stakes demand them, evals that run in CI, and observability so you can see exactly what an agent did and why. We work in the same stack you deploy in.

Where we say noNo agent goes to production without a regression suite and a kill switch. If a deterministic script would do the job, we write the script.

Deliverables

  • Agent architecture and tool design
  • Retrieval and knowledge pipelines
  • Evaluation harnesses and regression suites
  • Human-in-the-loop review interfaces
  • Model routing, cost and latency budgets
  • MCP servers and internal tool integration

StackTypeScript · Python · MCP · LangGraph · OTel

03Plumbing worth trusting

Data Engineering

Pipelines, warehouses, and contracts that make the numbers dependable — the unglamorous foundation every agent and dashboard stands on.

Most AI projects stall on data, not models. We build the ingestion, modeling, and quality layers first: typed schemas, tested transformations, lineage you can trace, and cost you can forecast. Then the interesting work has somewhere solid to stand.

Where we say noWe will not point an agent at an untested warehouse and call the output insight. Contracts and tests land before the demo.

Deliverables

  • Ingestion and streaming pipelines
  • Warehouse and lakehouse modeling
  • Data quality tests and contracts
  • Vector stores and embedding pipelines
  • Reporting and analytics interfaces
  • Migration off brittle legacy jobs

Stackdbt · Postgres · DuckDB · Airflow · pgvector

04Interfaces with a point of view

UX / UI Design

Research, product design, and design systems for software people actually have to use every day — including the hard, dense, expert-facing kind.

We design the interface and build it. Systems come out as real components with real tokens, not a static file handed over a wall. Accessibility, performance, and the awkward edge states are part of the design, not a follow-up ticket.

Where we say noWe don't ship a design system as a slide deck. If it isn't in code with tokens, it isn't a system.

Deliverables

  • Discovery, research, and service mapping
  • Product and interaction design
  • Design systems and component libraries
  • Prototypes that run in the browser
  • Accessibility and performance passes
  • Brand systems and art direction

StackReact · Next.js · Tailwind · Figma · Storybook

05Production and streaming

Multi-Media

Video, motion, audio, and live streaming platforms — customizable event experiences built fast and economically instead of from scratch.

EMME/STREAM began as a suite of pre-built features for bespoke virtual events: everything a cookie-cutter platform won't do, without the cost of a ground-up build. The same production capability carries into product film, motion systems, and interactive media.

Where we say noWe won't rebuild a streaming stack you could rent. The custom work goes where the event is actually different.

Deliverables

  • Custom streaming and event platforms
  • Video production and post
  • Motion design and title systems
  • Sound design and audio production
  • Interactive and kiosk media
  • Content pipelines and asset management

StackHLS · WebRTC · After Effects · Ableton

Not sure which
of these you need?

Neither are most people on the first call. Describe the problem and we will tell you which practice it belongs to — including when the answer is none of them.

Start a conversation