At CanJam NYC 2026 this past weekend, Grell’s latest design, the Grell OAE2 open back headphones, made their U.S. public debut. The $599 model builds on the original OAE1 and continues Grell’s pursuit of tonal accuracy, mechanical precision, and long-term listening comfort. The open back over ear design incorporates a newly optimized dynamic driver and an acoustically refined housing intended to improve airflow and project a more speaker like soundstage presentation in front of the listener.

If you’ve spent enough time walking the halls at hi-fi shows, you know the routine. Every year a handful of startups promise they’ve reinvented the wheel and that what you’re about to hear will change everything you thought you knew about speakers or headphones. You nod politely, sit down, listen for a few minutes, and try not to roll your eyes when the demo playlist inevitably lands on Diana Krall, the Eagles, or Norah Jones.

But when the name on the badge is Axel Grell, you stop joking around and actually pay attention.

Grell is hardly another “trust us, it’s revolutionary” startup voice on a crowded show floor. Before launching his own brand, the veteran headphone engineer spent decades at Sennheiser, where he was responsible for developing some of the most respected high end headphones of the past few decades. Small German company. You might have heard of it. They also make a decent strudel.

How the Grell OAE2 Tries to Move the Soundstage Out of Your Head

One of the biggest limitations of traditional headphones is the so called “in head” effect. Because the drivers sit millimeters from the ear and fire directly into the ear canal, most headphones create a listening perspective where instruments appear to originate from inside the listener’s head rather than from a believable space in front of them. While open back designs can widen the presentation and improve air and separation, they rarely change the fundamental geometry of how the sound reaches the ear. The result is often a presentation that feels spacious but still anchored inside the listener’s skull rather than resembling the externalized imaging produced by loudspeakers.

The Grell OAE2 open back headphones were engineered to address that specific limitation. Instead of following the conventional layout where the driver points straight into the ear canal, Axel Grell designed the acoustic structure so the output interacts more deliberately with the outer ear before entering the ear canal. This approach allows the pinna and surrounding ear structures to contribute to spatial cues in a way that more closely resembles how we hear speakers in a room.

In speaker listening, sound reaches the ear only after interacting with the head, shoulders, and outer ear, creating small timing, phase, and tonal variations that the brain uses to interpret direction, distance, and placement. By preserving more of those interactions inside the headphone structure, the OAE2 attempts to shift the listening perspective forward so that instruments appear positioned in front of the listener rather than inside the head.

The goal is not to artificially exaggerate soundstage width or create gimmicky spatial effects, but to maintain stable imaging, natural treble perception, and controlled low frequency behavior while presenting music in a way that resembles nearfield loudspeaker listening.

For listeners accustomed to the traditional headphone presentation, the perspective can initially feel unfamiliar, but the intention is that the brain adapts to the spatial cues over time, making the presentation feel more natural and less fatiguing during long listening sessions.

German Engineering, Replaceable Parts, and None of That Disposable Headphone Nonsense

Beyond the acoustic design, the Grell OAE2 reflects Axel Grell’s long standing belief that premium headphones should be built to last. Instead of chasing short product cycles, the design emphasizes durability, serviceability, and long term ownership. In other words, the opposite of the sealed plastic approach that dominates much of the modern headphone market.

At the center of the OAE2 is a 40 mm wideband dynamic driver built around a bio cellulose diaphragm, paired with a carefully tuned damping system. Part of that system includes a precision manufactured stainless steel acoustic mesh produced in Germany, which helps regulate airflow and maintain consistent driver behavior. The goal is controlled acoustics rather than brute force tuning, supporting the headphone’s spatial presentation without introducing unwanted resonances or instability.

Construction follows a modular all metal architecture with replaceable components that can be serviced if parts wear out. The idea is simple: headphones should not become disposable because one component fails. Connectivity is equally straightforward. The OAE2 ships with two detachable 1.8 m cables, including a 3.5 mm single ended cable and a 4.4 mm balanced cable, along with a screw on 3.5 mm to 6.3 mm adapter for traditional headphone amplifiers and a protective carry case.

From a technical standpoint, the OAE2 remains close to the OAE1 but with small refinements. The circumaural open back design uses a dynamic transducer rated from 12 Hz to 34 kHz within ±3 dB, extending from 6 Hz to 46 kHz at -10 dB. Nominal impedance is 38 ohms with 100 dB sensitivity at 1 kHz (1 VRMS), making it compatible with portable players while still benefiting from a capable amplifier. Total harmonic distortion is rated at 0.05 percent at 1 kHz and 100 dB, and the headphone weighs 378 g (13.3 oz) without the cable attached. Slightly heavier than the OAE1 by three grams. German engineering apparently does not skip arm day.

The Germans Ran the Numbers. Now We Listen.

Grell kept things refreshingly simple at CanJam NYC 2026. No $30,000 source chain, no mystical demo playlist, and no attempt to overwhelm people with exotic gear. Just three pairs of the Grell OAE2, a few source devices, and a small stack of EarMen ST-Amp headphone amplifiers priced around $400.

The setup was about as straightforward as it gets for a show floor demo. The EarMen ST-Amp offers both single ended and balanced output options, rated at 0.5W into 32 ohms (4V) single ended and 1.85W into 32 ohms (7.75V) balanced, which is more than enough for a 38 ohm headphone like the OAE2. In other words, plenty of clean power but nothing exotic that might artificially inflate the listening experience.

Even the music selection avoided the usual trade show clichés. There were no sacred audiophile demo tracks looping endlessly in the background. Attendees could simply plug in their own phone and listen to whatever they wanted. Which I appreciated. I even spent a little time surfing through German tracks on the playlist. My German is… limited. Although if knowing Yiddish counts as partial credit, I was doing just fine.

A short run through Deadmau5, Daft Punk, and Aphex Twin was enough to get my attention. The OAE2 leans toward a neutral presentation without obvious boosts or dips across the spectrum. Clean, but not the sterile kind of clean that some German designs fall into. Bass is not tuned for exaggerated punch, but the definition and speed are excellent, especially with electronic music that depends on tight timing. The top end is equally well behaved. Detailed and open without any glare or hardness.

Then came the part that really mattered. Soundstage. The presentation is clearly wide, but not absurdly wide in the artificial sense. Think more East River to the IAC Building on the West Side Highway wide, not across the Hudson into New Jersey for chili dogs at Hiram’s wide. More important than width, however, is placement in front of the listener. If Axel Grell’s goal was to push the image outside the head and create what I would call a “nearface” listening perspective about 6 to 10 inches in front of you, the OAE2 largely succeeds.

Switching over to vocals confirmed that impression. Tracks from Amy Winehouse, Billie Holiday, Bjork, and Belinda Carlisle showed the same spatial behavior. Voices did not collapse into the center of the skull as they often do with headphones. Some appeared slightly closer, others further back, but the image remained focused and stable. Never diffuse. Always locked in dead center. And frankly, that was pretty impressive.

The Bottom Line

The Grell OAE2 stands out for one reason: it actually delivers on the promise of moving the soundstage out of your head and placing it slightly in front of the listener. The effect is not gimmicky or exaggerated. Instead, it feels closer to a nearfield speaker presentation with stable imaging and natural placement. Sonically the tuning leans neutral with clean treble, fast and well controlled bass, and no obvious peaks designed to impress in a quick demo.

Comfort was also encouraging. Clamping force is moderate, the headband is well padded, and the overall build quality feels appropriate for a headphone expected to retail for $599 / £499 / €499. Based on our first listen at CanJam NYC 2026, Axel Grell’s latest design shows real promise. A full review is coming later this month once we spend more time with a production unit.

For more information: grellaudio.com

Related Reading:

The hype for Gang of Dragon, the debut game from Nagoshi Studio, may already be getting derailed. According to a Bloomberg report, Chinese tech giant NetEase is going to stop financing Nagoshi Studio starting in May. Bloomberg confirmed the news with the studio’s employees and a NetEase spokesperson.

The report explained that NetEase decided to cut funding to Nagoshi Studio, which was founded in 2021 by Yakuza franchise creator Toshihiro Nagoshi, after finding out the studio needed $44.4 million to complete the project. Bloomberg reported that Nagoshi Studio is trying to find new sponsors but hasn’t had any success so far. The report also added that the studio can continue the project on its own, but would be responsible for paying NetEase for any associated costs to hold onto the brand or assets.

While Nagoshi Studio may have been working on Gang of Dragon since the studio’s creation, the general public got a better look at the title through a trailer announcement during The Game Awards 2025. The action-adventure game set in Tokyo would star Ma Dong-Seok, a South Korean actor who starred in Train to Busan and Marvel’s Eternals. As of now, Nagoshi Studio might be at risk of joining other casualties stemming from NetEase’s executive decisions, like when the tech giant decided to shut down Ouka Studio in 2024.

Coffee is the original office biohack and the nation’s most popular productivity tool. As we lose sleep to the changeover to daylight saving time, the caffeine-addicted WIRED Reviews team is writing about our favorite coffee brewing routines and devices that’ll keep us alert and maybe even happy in the morning. Today, operations manager Scott Gilbertson expounds on the perfect simplicity of the moka pot. In the days after, we’ll add other Java.Base stories about other WIRED writers’ favorite brewing methods.

Years of travel and a love of repair has given me a special appreciation for simple devices. A pen and paper is still the simplest way to write. A cast-iron pan is the simplest way to cook. And a moka pot is the simplest way to brew coffee.

What I love about the moka pot isn’t just the results I get from it. I do love the flavor, especially when paired with a nice dark, chocolatey, smokey roast, but the moka pot is about more than flavor. It’s also about ingenious simplicity and a design that has lasted nearly a century.

Simple Beginning

Photograph: Scott Gilbertson

The moka pot’s exact origins depend on who you ask, but it was first manufactured and popularized by an aluminum manufacturer named Alfonso Bialetti and his son Renato, who started mass-producing them the same year. Today, Bialetti Industries still makes the Moka Express. The iconic logo image of a short, squat, heavily mustached man is indeed based on Bialetti himself.

If you want some idea of Renato Bialetti’s commitment to the device that made him famous, consider that when he died in 2022, his ashes were interred in a large moka pot. He isn’t the only one who revered the design. The moka pot is featured in museums around the world, including the Museum of Modern Art. Its iconic octagonal shape makes it one of the most recognized coffee brewers in the world.

The moka pot is a pressure-driven stovetop (or campfire top, though this requires close attention) coffee brewer that works something like a percolator. The Moka Express consist of four parts, split into two chambers. The bottom is the water reservoir which heats up on the stove. Into this, you put the brewing basket which holds your grounds. The top consists of a long tube in the center of a holding chamber. On the bottom of the top piece, there’s a metal filter ringed by a rubber (or silicone on some models) gasket. The top and bottom screw together.

As the water heats it passes upwards, through the basket of grounds, and eventually out of the tube. The extraction sits above the grounds and the metal filter keeps everything in place. It’s ingeniously simple.

“When you get a demo and something works 90% of the time, that’s just the first nine.” — Andrej Karpathy

The “March of Nines” frames a common production reality: You can reach the first 90% reliability with a strong demo, and each additional nine often requires comparable engineering effort. For enterprise teams, the distance between “usually works” and “operates like dependable software” determines adoption.

The compounding math behind the March of Nines

“Every single nine is the same amount of work.” — Andrej Karpathy

Agentic workflows compound failure. A typical enterprise flow might include: intent parsing, context retrieval, planning, one or more tool calls, validation, formatting, and audit logging. If a workflow has n steps and each step succeeds with probability p, end-to-end success is approximately p^n.

In a 10-step workflow, the end-to-end success compounds due to the failures of each step. Correlated outages (auth, rate limits, connectors) will dominate unless you harden shared dependencies.

Define reliability as measurable SLOs

“It makes a lot more sense to spend a bit more time to be more concrete in your prompts.” — Andrej Karpathy

Teams achieve higher nines by turning reliability into measurable objectives, then investing in controls that reduce variance. Start with a small set of SLIs that describe both model behavior and the surrounding system:

• Workflow completion rate (success or explicit escalation).

• Tool-call success rate within timeouts, with strict schema validation on inputs and outputs.

• Schema-valid output rate for every structured response (JSON/arguments).

• Policy compliance rate (PII, secrets, and security constraints).

• p95 end-to-end latency and cost per workflow.

• Fallback rate (safer model, cached data, or human review).

Set SLO targets per workflow tier (low/medium/high impact) and manage an error budget so experiments stay controlled.

Nine levers that reliably add nines

1) Constrain autonomy with an explicit workflow graph

Reliability rises when the system has bounded states and deterministic handling for retries, timeouts, and terminal outcomes.

• Model calls sit inside a state machine or a DAG, where each node defines allowed tools, max attempts, and a success predicate.

• Persist state with idempotent keys so retries are safe and debuggable.

2) Enforce contracts at every boundary

Most production failures start as interface drift: malformed JSON, missing fields, wrong units, or invented identifiers.

• Use JSON Schema/protobuf for every structured output and validate server-side before any tool executes.

• Use enums, canonical IDs, and normalize time (ISO-8601 + timezone) and units (SI).

3) Layer validators: syntax, semantics, business rules

Schema validation catches formatting. Semantic and business-rule checks prevent plausible answers that break systems.

• Semantic checks: referential integrity, numeric bounds, permission checks, and deterministic joins by ID when available.

• Business rules: approvals for write actions, data residency constraints, and customer-tier constraints.

4) Route by risk using uncertainty signals

High-impact actions deserve higher assurance. Risk-based routing turns uncertainty into a product feature.

• Use confidence signals (classifiers, consistency checks, or a second-model verifier) to decide routing.

• Gate risky steps behind stronger models, additional verification, or human approval.

5) Engineer tool calls like distributed systems

Connectors and dependencies often dominate failure rates in agentic systems.

• Apply per-tool timeouts, backoff with jitter, circuit breakers, and concurrency limits.

• Version tool schemas and validate tool responses to prevent silent breakage when APIs change.

6) Make retrieval predictable and observable

Retrieval quality determines how grounded your application will be. Treat it like a versioned data product with coverage metrics.

• Track empty-retrieval rate, document freshness, and hit rate on labeled queries.

• Ship index changes with canaries, so you know if something will fail before it fails.

• Apply least-privilege access and redaction at the retrieval layer to reduce leakage risk.

7) Build a production evaluation pipeline

The later nines depend on finding rare failures quickly and preventing regressions.

8) Invest in observability and operational response

Once failures become rare, the speed of diagnosis and remediation becomes the limiting factor.

• Emit traces/spans per step, store redacted prompts and tool I/O with strong access controls, and classify every failure into a taxonomy.

• Use runbooks and “safe mode” toggles (disable risky tools, switch models, require human approval) for fast mitigation.

9) Ship an autonomy slider with deterministic fallbacks

Fallible systems need supervision, and production software needs a safe way to dial autonomy up over time. Treat autonomy as a knob, not a switch, and make the safe path the default.

• Default to read-only or reversible actions, require explicit confirmation (or approval workflows) for writes and irreversible operations.

• Build deterministic fallbacks: retrieval-only answers, cached responses, rules-based handlers, or escalation to human review when confidence is low.

• Expose per-tenant safe modes: disable risky tools/connectors, force a stronger model, lower temperature, and tighten timeouts during incidents.

• Design resumable handoffs: persist state, show the plan/diff, and let a reviewer approve and resume from the exact step with an idempotency key.

Implementation sketch: a bounded step wrapper

A small wrapper around each model/tool step converts unpredictability into policy-driven control: strict validation, bounded retries, timeouts, telemetry, and explicit fallbacks.

def run_step(name, attempt_fn, validate_fn, *, max_attempts=3, timeout_s=15):

    # trace all retries under one span

    span = start_span(name)

    for attempt in range(1, max_attempts + 1):

        try:

            # bound latency so one step can’t stall the workflow

            with deadline(timeout_s):

                out = attempt_fn()

# gate: schema + semantic + business invariants

            validate_fn(out)

            # success path

            metric(“step_success”, name, attempt=attempt)

            return out

        except (TimeoutError, UpstreamError) as e:

            # transient: retry with jitter to avoid retry storms

            span.log({“attempt”: attempt, “err”: str(e)})

            sleep(jittered_backoff(attempt))

        except ValidationError as e:

            # bad output: retry once in “safer” mode (lower temp / stricter prompt)

            span.log({“attempt”: attempt, “err”: str(e)})

            out = attempt_fn(mode=”safer”)

    # fallback: keep system safe when retries are exhausted

    metric(“step_fallback”, name)

    return EscalateToHuman(reason=f”{name} failed”)

Why enterprises insist on the later nines

Reliability gaps translate into business risk. McKinsey’s 2025 global survey reports that 51% of organizations using AI experienced at least one negative consequence, and nearly one-third reported consequences tied to AI inaccuracy. These outcomes drive demand for stronger measurement, guardrails, and operational controls.

Closing checklist

• Pick a top workflow, define its completion SLO, and instrument terminal status codes.

• Add contracts + validators around every model output and tool input/output.

• Treat connectors and retrieval as first-class reliability work (timeouts, circuit breakers, canaries).

• Route high-impact actions through higher assurance paths (verification or approval).

• Turn every incident into a regression test in your golden set.

The nines arrive through disciplined engineering: bounded workflows, strict interfaces, resilient dependencies, and fast operational learning loops.

Nikhil Mungel has been building distributed systems and AI teams at SaaS companies for more than 15 years.

Welcome to the VentureBeat community!

Our guest posting program is where technical experts share insights and provide neutral, non-vested deep dives on AI, data infrastructure, cybersecurity and other cutting-edge technologies shaping the future of enterprise.