Flywheel AI

What do they actually do

Flywheel AI sells a retrofit kit and software that lets contractors operate existing excavators remotely. Their hardware attaches to the machine, exposes the controls to a remote console, and streams multi‑camera video and machine telemetry to a web interface so a trained operator can drive from off‑site useflywheel.ai, YC profile, Launch HN.

Today the company focuses on teleoperation deployments and paid pilots with contractors, emphasizing installs on customers’ current fleets rather than buying new equipment. The immediate value is safer operation and staffing flexibility (e.g., one operator can cover multiple sites without traveling) useflywheel.ai, LinkedIn announcement.

While machines are teleoperated, the system records synchronized video and control inputs to build a dataset for training assisted‑automation features. Flywheel has publicly released an “Excavator 100” dataset with roughly 100 hours of multi‑camera plus joystick data collected on real sites, underscoring the teleop‑first, autonomy‑later roadmap Launch HN, Hugging Face dataset, YouTube dataset video.

Who are their target customer(s)

• Small-to-mid-size general contractors running many small sites: They struggle to staff each job with qualified excavator operators and lose time shuttling crews. They need machines working without adding more on‑site headcount.
• Excavation subcontractors and yard operators with fleets: Utilization drops when operators are idle or traveling, and experienced operators are costly and scarce. They want one operator to cover multiple machines or sites to reduce labor pressure.
• Equipment fleet managers and rental companies: Hand‑offs between sites increase damage/theft risk and create logistics delays. Remote control/monitoring can cut site visits and speed turnarounds.
• Large contractors doing repetitive excavation (trenching, grading): They need predictable, repeatable outcomes on boring, error‑prone tasks and face shortages of skilled operators. Assisted automation can standardize results and let supervisors oversee more machines.
• Owners of hazardous, remote, or regulated sites (remediation, mining, disaster response): Putting people on‑site raises safety, compliance, and travel costs; some locations are unsafe for routine work. Remote operation reduces exposure while keeping projects moving.

How would they acquire their first 10, 50, and 100 customers

• First 10: Founder-led local pilots: personally install retrofit kits on nearby contractor machines, run short paid pilots with reduced install fees, and document before/after safety and uptime to justify purchase and refine the install/checklist useflywheel.ai, YC profile.
• First 50: Convert pilot users into reference accounts and referrals with a small discount, hire 1–2 regional field engineers for installs/demos, and sell packaged pilot offers to excavation subs and rental yards using a standardized install checklist, pricing template, and two-page case study.
• First 100: Scale through channels and enterprise pilots: partner with rental houses and regional dealers so retrofit + teleop is offered at rental/deployment, stand up an install team and CS playbook for operator onboarding, and use published datasets/case studies to close multi-machine, multi-site contracts Launch HN.

What is the rough total addressable market

Top-down context:

The global excavator market is large and mature—estimated around $81B in 2024—implying a sizable installed base of machines that could be retrofitted for teleoperation and future assisted autonomy Grand View Research. Asia-Pacific holds the largest regional share, with North America and Europe comprising a substantial remainder IMARC.

Bottom-up calculation:

If Flywheel targets ~100k excavators in North America and Europe (a subset doing repetitive construction tasks and open to retrofits), and charges ~$20k per retrofit plus ~$6k/year software, the near-term hardware TAM is ~$2.0B with ~$600M in recurring annual software revenue. This represents a focused beachhead within a much larger global fleet.

Assumptions:

• Average new excavator ASP ~$150k–$250k (typical ranges: mini ~$30k–$110k; medium ~$190k–$300k; large ~$300k–$500k+) Quipli, Five Star Equipment.
• Installed base inferred as multiple years of sales; NA+EU together account for a sizable share (Asia-Pacific ~43% in 2024) IMARC.
• Adoption focus on ~100k machines initially (subset of NA+EU fleets in repeatable tasks) at ~$20k hardware and ~$6k/year software; actual pricing and attach rates may vary.

Who are some of their notable competitors

• Built Robotics: Aftermarket autonomy kits for dozers and excavators focused on trenching and other repeatable tasks; overlaps with Flywheel’s long‑term autonomy goal and retrofit path Built Robotics, press.
• Caterpillar (Cat Command): OEM teleoperation/remote control options for Cat machines, competing on factory integration, dealer support, and brand trust Cat Command, press release.
• SafeAI: OEM‑agnostic autonomy and retrofit conversions for heavy equipment fleets; competes on autonomy retrofits and fleet modernization coverage, partnership news.
• Komatsu: OEM with teleoperation and semi‑autonomous demonstrations and global dealer channels; attractive for buyers preferring factory‑backed solutions demo.
• Phantom Auto (historical): Built a cloud teleoperation platform for logistics/industrial vehicles; shutdown in 2024 highlights both demand and the scale-up challenges in teleop businesses TechCrunch.