Building in the open.

MachBeat is under active development, and we’d rather be honest about it. Here is the roadmap, the real build status of every layer, and answers to the questions engineers actually ask.

Book a Free Pilot → About the company

Roadmap

From working pilot to enterprise platform.

The plan is sequenced so early revenue funds the next layer — and so we don’t build scale features before there are customers who need them.

Phase 0 · Now

Pilot-ready

End-to-end device-online, TLS on transport, idempotent ingest, health checks and email alerts.

→ Unlocks: first paid pilot

Phase 1 · 1–2 quarters

Design partners

Durable streaming buffer, time-series store and windowed AI to break the per-packet ceiling; deeper hierarchy and roles; reports & export; envelope + cepstrum DSP.

→ Unlocks: 3–5 factories, real case studies

Phase 2 · 2–3 quarters

Multi-tenant SaaS

SSO / SCIM, billing + per-asset metering, mobile app v1, observability & DR, and the first unsupervised-ML anomaly tier.

→ Unlocks: self-serve mid-market

Phase 3 · 3–4 quarters

Scale

Autoscale to 100k+ devices, the 3D Vision twin, CMMS/ERP connectors, and white-label.

→ Unlocks: enterprise pilots

Phase 4 · 4–6 quarters

Enterprise & compliance

SOC 2 Type II + ISO 27001, deep predictive models with MLOps, multi-cloud, and an OEM channel.

→ Unlocks: Fortune-500 contracts

Honest build status

What runs today, what’s partial, what’s next.

The same matrix the founding team works from. No layer is claimed as finished before it is.

Layer	Built	Partial	Not started
Edge	Vibration node, DSP, buffering, OTA hook, provisioning	Industrial gateway, cert identity, multi-sensor	Secure-boot hardware, enclosure
Cloud	MQTT ingest, device registry, database, REST + Swagger, WebSocket, RBAC, audit	Tenancy depth, TLS	Streaming buffer, time-series, autoscale, connectors
AI	L0 heuristics, FFT, bearing, health, RUL, cost, explainable evidence	Envelope / cepstrum	L2/L3 ML, MLOps, retraining
SaaS	Multi-tenant core, roles, billing tiers	Hierarchy depth, full role set	SSO/SCIM, billing engine, white-label
Apps	Mission-control web, 2D digital twin, explainable AI	Reports / export	Mobile, 3D twin
Security	Argon2, JWT + refresh, RBAC, audit, rate-limit	TLS, device certs	SOC 2 / ISO, signed OTA
Ops	Docker, CI, compose	Health checks	Kubernetes, observability, DR, backups

FAQ

Questions engineers actually ask.

What machines can MachBeat monitor?

Rotating industrial assets — motors, pumps, compressors, fans and gearboxes — are the primary focus. Fit for a specific machine is confirmed during the pilot.

What does the sensor measure today?

Triaxial vibration and temperature with operating-state context. Additional parameters and an industrial multi-sensor gateway are on the roadmap.

Do I need existing sensor history for the AI to work?

No. The L0 heuristic and L1 DSP tiers work cold-start, with no failure history. The unsupervised and supervised tiers deepen the diagnosis as your machines accumulate data.

How does data reach the platform?

The edge node streams over MQTT into the MachBeat backend. Transport TLS and per-device certificates are being rolled in; a local-network path is used for lab and early pilots.

How are alerts generated?

Trends are compared against recent baselines and evaluated by the AI engine. Early indications highlight deviation with root cause and a recommended action — they are investigative cues, not guaranteed failure predictions.

Is it secure enough for enterprise?

The app has role-based access, JWT auth and a full audit log today. TLS and device certificates are in progress, and a mapped path to SOC 2 Type II and ISO 27001 is on the roadmap before the first large-enterprise deployment.

Can it integrate with our CMMS/ERP?

A REST API with Swagger docs is available today. Webhooks and CMMS/ERP connectors (e.g. SAP, Maximo) are being added — integration scope is discussed during the pilot.

How long does a pilot take?

It depends on the site and asset. During the pilot we mount, connect, collect a baseline window, and review signals with your team.

Want the technical deep-dive?

We’re happy to walk an engineering team through the architecture, the AI ladder, and exactly what runs today.

Book a Free Pilot → Talk to an Engineer