AI Imagineering

Spaces and objects that listen back.

We're an AI Imagineering studio. Voice is where most of the work ships today: a rotary phone in a theatre foyer, a room that listens back. The same stack runs museum-scale builds, custom Claude tools, and AI integrations.

Eindhoven · Strijp-S Hardware, software, AV Claude-native, model-agnostic

Closer to imagineers than to consultants.

We design the experience, we build the hardware, we write the AI. Experience design and deep-tech engineering sit next to each other in the studio. The room, the character, the way an interaction feels in your hands. The systems behind that, the ones that hold up under load and survive the edge cases nobody saw coming.

Eindhoven, on Strijp-S, is where this kind of work belongs. Rooted in the Philips heritage of the city: people who fabricate, prototype, and ship in the same week. The studio is small on purpose. The team that scopes the project is the team that builds it.

If a Reckon Day surfaces it, the studio can build it.

Voice is where most of the work ships.

Voice AI is the lane where the studio has the most miles. The rotary phone we built for Parktheater Eindhoven ran 26 days unattended and held 613 real conversations with theatre visitors, with no human moderation behind the scenes. The engine inside the pedestal is Switch Board AI, our voice AI product, named and shipped from inside Reckon. Studio.

Alongside Switch Board AI, the studio runs two other products: Reck Connect, our station-and-satellite cockpit for working with Claude, and Reck Reader, a small hardware reading device paired to Chrome over Bluetooth. We also build custom Claude tools, dashboards, agents, and integrations on a brief. The same craft that builds products under our own name is the craft we apply on client work.

  • Voice AI installations. Objects, characters, kiosks, rooms. Voice in, story out, no app required.
  • Feedback systems. The P.I.M. shape, on real hardware. Honest qualitative input from real visitors, into a dashboard your team can act on.
  • Museum-scale and experience installations. Rooms, sequences, immersive sets. End-to-end design, fabrication, and integration.
  • Audio-tour and exhibit triggers. Voice AI installed on existing exhibits, cueing lights, sound, and movement on the venue's own hardware. Portable demo available for site visits.
  • Custom Claude/AI tools and integrations. Internal apps, dashboards, agents, integrations. Small surface, sharp edge, real adoption.
  • 360-film, immersive video, and VR integrations. When the medium is video or virtual, we wire AI through it.
  • Audio, lighting, AV pipelines. When the experience leaves the screen, we run the cabling too.

Beyond voice, on existing hardware.

The studio runs a portable Voice AI audio tour that installs on existing exhibits. A visitor speaks, the exhibit answers, and the same voice engine can cue lights, trigger sounds, and drive movement on the mechanisms the venue already owns. It turns a static object into a conversation, with most of the gear already on the floor. Built, demoed, ready to install.

RFID-aware environments run on the same engine, with physical objects knowing each other and the room knowing who's holding what. Sound and lighting patched into the voice loop, not static DMX. These pieces are how a voice-first studio gets into whole experiences without asking a venue to throw out what it already has.

Toward animatronics and museum revamps.

The longer arc points at animatronics that respond to who's standing in front of them, full museum-scale installations, and venue-wide builds where voice, RFID, sound, and lighting are one integrated system. Some of this is in the workshop. Some is on the bench. We'll show it when it's live.

We're not pretending it's already done. Honestly, voice AI is what runs today, the rest is the road. If you have a brief that lives somewhere on that road, the conversation is welcome.

The 5 R's, applied to building instead of teaching.

Same cycle, different shape. We recognize what the organisation actually does and where AI fits. We reckon with the constraints (legal, technical, cultural, physical). We refine in short sprints with working code, never PowerPoint. We reflect alongside you to keep the build honest. And we run with the team that owns it after we leave.

  • Discovery sprint, two weeks. Honest reckoning with workflows, constraints, and feasibility. Output: a one-page brief and a clear go or no-go.
  • Build sprint, four to eight weeks. Working software each week, demoed to the people who'll use it. No big-bang launches.
  • Run, ongoing. Hand-off to your team or a small monthly retainer. Honest end-of-life when the thing has run its course.

Same five steps we run for ourselves, we run with you. Claude-native today, but we're not religious about it. The right model for the job, including local models when privacy or latency demands.

Cultural venues, scale-ups, agencies, and museum partners.

The deepest work has been with cultural venues and theatres where the bar for trust, privacy, and physical reliability is high. We also work with scale-ups that need a small engineering team for a quarter rather than a year, and with creative agencies who want voice AI inside experiences they're already building.

If your context isn't on that list, write us anyway. The honest tell is whether the problem fits the cycle, not whether the sector matches a case study.

Sharp answers to the practical questions.

Are you an agency, a consultancy, or a software house?

None exactly. We're a small AI engineering and experience-design studio. Closer to a boutique product team than a consultancy. We build working software and physical objects, not slide decks. If you need a partner who can also stand in front of a camera and explain it, that's also us.

Why hire a small studio instead of a bigger agency?

Most agencies split conceptual design from hardware execution across two teams, sometimes two firms. The meetings between them tend to eat around 70% of the budget. Reckon. does both crafts under one roof: the people who scope the experience are the same people who write the AI and build the hardware. Fewer translation layers, more making. We bring in trusted collaborators for specific shapes (industrial design, fabrication, additional engineering hands) when scope demands it.

Do you do animatronics today?

Not as a shipped capability yet. It's a direction we're moving toward, anchored in our voice AI work and our Eindhoven fabrication network. If your brief is animatronics-led, we'll be honest about where we are versus where we're going. If your brief is voice-led with movement as a layer on top, that's something we can plan today.

What about RFID-aware environments and sensing rooms?

In flight. We have RFID and sensor prototypes on our bench, and the same Claude-native stack that powers our voice work extends naturally into spaces where objects know each other. The first commercial deployment of this work will be public when it ships.

Do you do data infrastructure projects?

Only when they sit underneath an AI use case we're building. We're not a generalist data team. If your problem is a five-person data engineering project, we'll point you to people who specialise in that.

How does GDPR work in your builds?

From day one. Data minimisation, anonymisation, retention windows, and processing-region awareness are all part of the discovery sprint. Cultural-sector clients have signed off our setup. We can share the architecture under NDA.

Can a Reckon Day lead to a studio engagement?

Often. A Reckon Day is the cheapest way to find out whether a build is worth doing. About a third of our studio engagements start as a single Reckon Day with a team that surfaces the right idea.

Got something to build?

Tell us what you want built.

Voice installations, museum work, custom Claude tools, experience triggers, feedback systems. lets@reckon.works.