1. Headline & intro
AI hardware keeps crashing into the same wall: clever demos, confused users, and dead products. Era is betting the problem isn’t the gadgets themselves, but the missing software layer underneath them.
By quietly raising $11 million and staying out of the hardware spotlight, the startup is trying to become the “intelligence OS” for whatever comes after the smartphone: pins, pendants, smart jewelry, ambient objects. In this piece we’ll look at what Era is actually building, why the platform play is far smarter than shipping yet another AI device, and what this could mean for European makers, regulators and consumers.
2. The news in brief
According to TechCrunch, New York–based startup Era has raised a total of $11 million to build a software platform for AI-powered gadgets.
The financing includes a $9 million seed round led by Abstract Ventures and BoxGroup, with participation from Collaborative Fund and Mozilla Ventures, on top of a $2 million pre-seed from Topology Ventures and Betaworks. Angels include Flickr co-founder Caterina Fake and several founders from the AI and hardware world.
Era, founded in 2025 by CEO Liz Dorman, CTO Alex Ollman and CPO Megan Gole, provides a cloud-based orchestration layer that connects hardware devices to more than 130 large language models from over 14 providers. The company does not plan to build its own devices; instead, it offers tools for others to add voice interfaces, agentic behavior and multimodal intelligence to form factors like glasses, jewelry and home speakers.
TechCrunch reports that Era has already distributed developer kits to artists and makers, who have built experimental devices such as stock-monitoring trinkets and air-quality gadgets.
3. Why this matters
The AI gadget space is littered with wreckage. Humane was sold to HP, Rabbit has largely gone silent, and countless crowdfunded “AI wearables” never make it beyond early adopters. Most of these companies tried to do everything at once: new hardware, new software stack, new interaction model, new distribution.
Era is taking the opposite approach: become infrastructure.
By positioning itself as the intelligence layer rather than the device, Era sidesteps the riskiest part of the stack: expensive hardware bets in a market that doesn’t yet know what it wants. Its customers can experiment with dozens of weird form factors and niche use cases; Era just needs some of them to stick.
The orchestration angle is important. Routing across 130+ models and 14 providers isn’t a gimmick – it attacks three real problems for small hardware teams:
- Cost and latency optimisation: using a cheap model for simple queries and a heavier one for complex tasks.
- Resilience to outages and regulation: if one provider changes terms or goes down, devices keep working.
- Feature diversity: mixing models that are better at code, vision, summarisation, or local languages.
For developers, this could turn AI gadgets from a research project into something shippable without hiring a full ML team.
The risk is classic platform risk: if Apple, Google, or the major clouds decide to offer similar orchestration tightly integrated into their ecosystems, Era becomes a feature, not a company. Its bet is that incumbents will stay focused on phones and PCs, leaving niche, experimental hardware to independents.
4. The bigger picture
Era fits into at least three broader trends.
First, the shift from apps to agents. Over the last two years we’ve seen OpenAI, Anthropic and others push “agentic” frameworks where you state a goal and the system figures out the steps. Applying that idea to hardware means the device isn’t just a voice remote for apps, but an autonomous actor: schedule a trip, call a cab, order groceries. Era explicitly talks about replacing the app layer with an “intelligence layer” – that’s the same direction, but expressed in silicon instead of browser tabs.
Second, the Cambrian explosion of form factors. TechCrunch’s description of experimental gadgets – souvenirs that talk about France, stock-quit-your-job meters, environmental sensors – highlights how cheap components and commoditised manufacturing have become. We’ve been here before: the early Android era, when every manufacturer tried a slightly different phone. The difference now is that the “OS” isn’t just managing hardware, it’s arbitrating between AI models and modalities.
Third, the rise of model routing as a business. From OpenRouter to Nvidia’s NIM and cloud providers’ model gardens, everyone wants to sit in the middle and decide which model handles which call. Era is that idea, specialised for devices with flaky connectivity, constrained compute and strict power budgets. It’s closer to an AI-focused version of AWS IoT or Azure IoT Hub than to yet another SDK.
If Era is right, we don’t end up with one iconic “AI gadget” like the iPhone, but a long tail of context-specific objects – many of them built by small brands, artists and industrial OEMs. Someone has to make that chaos manageable.
5. The European / regional angle
For Europe, AI gadgets are less about novelty and more about compliance, trust and sovereignty.
Ambient devices that constantly listen, see or infer context are walking GDPR headaches. Who is the controller? Where is the data stored? How are models trained? Era’s stated ambition to let users choose their memory and model providers in a privacy-preserving way could align surprisingly well with EU expectations – if it’s real, technically robust, and well documented.
Under the EU AI Act, many of the models Era routes to will be classified as general-purpose AI, with obligations on both providers and deployers. A platform like Era could either simplify compliance for European hardware makers (one integration, many models, centralised logging and consent flows) or complicate it (another opaque US middleman in the stack). How it handles data residency, logging, and opt-out mechanisms will be watched closely by privacy regulators in Germany, France and beyond.
On the opportunity side, this is catnip for Europe’s fragmented hardware ecosystem. From industrial players like Bosch and Siemens to smaller IoT manufacturers in Central and Eastern Europe, there is huge demand for vertical, specialised devices: factory wearables, logistics badges, smart city infrastructure. Most of these companies will never build their own AI stack. If Era offers solid on-prem or EU-cloud options, it could become a default choice for such projects.
The flip side: dependence on a US platform for yet another layer of the stack runs against Europe’s push for digital sovereignty. Expect European cloud providers and open-source consortia to eye the same space.
6. Looking ahead
Over the next 12–24 months, expect the AI gadget field to resemble early smartphone app stores: noisy, chaotic, and mostly forgettable – but with a few hits that quietly define expectations.
For Era, three milestones will determine whether it becomes infrastructure or trivia:
- Developer traction: can it win not just artists and hackers, but mid-size hardware OEMs who ship tens of thousands of units? Showcases are nice, but design wins pay the bills.
- Reliability in the messy real world: devices go offline, users shout over background noise, firmware doesn’t get updated. An orchestration platform that works beautifully in a lab can crumble in a kitchen or on a factory floor.
- Positioning against giants: if Google folds richer agentic APIs into Android, or Apple doubles down on its own on-device AI frameworks for wearables, Era must either integrate gracefully or shift toward markets the giants don’t care about (industrial, medical, education, creative tools).
Business-model-wise, usage-based pricing per token or per interaction looks obvious, but hardware makers hate unpredictable costs. Expect pressure for hybrid models: per-device licensing with usage caps, or white-label offerings for enterprises.
The biggest open question: will consumers accept another class of dedicated devices, or will the smartphone continue to absorb all useful AI interactions? Era’s fate is tied to the answer.
7. The bottom line
Era is making the smartest kind of bet in AI hardware: not on a single shiny gadget, but on the messy, heterogeneous future that comes after the hype. If AI gadgets become a real category, an independent intelligence layer could be as important as Android was for smartphones. If they don’t, this is an expensive science project.
The real question for readers is simple: what kind of AI interaction would be compelling enough for you to wear, carry or install a new device – instead of just using your phone?
