Supervised robot-operations OS for refineries that sends mixed ground robots into hot zones first, with auditable clearance before crews enter.

1. Fresh seed capital behind Shifters shows supervised ground robots are moving into field programs now, so software that operationalizes mixed robot teams can ride an active hardware budget line.
2. Shifters is explicitly building supervised autonomy, which means human-in-the-loop workload, approvals, and auditability are now central product problems rather than edge cases.
3. The first buyers are defense, security, and hazardous-environment teams where every avoided human entry has immediate safety and downtime value.
4. Manufacturing readiness and expansion across the U.S., Middle East, and Europe suggest operators will soon manage more robot programs across more sites, favoring a repeatable operating layer over bespoke services.

Catalyst. Shifters' financing and emphasis on supervised autonomy, hazardous missions, and manufacturing readiness suggest robot supply is arriving now, making deployment workflow software the gating layer for real field use.

The product is a vendor-neutral command and evidence layer that sits above existing UGVs, crawlers, and sensor payloads. It gives a human supervisor prebuilt mission playbooks for post-incident perimeter sweeps, gas-reading collection, confined-space scouting, and valve or breaker verification, with autonomy envelopes that define what each robot may do without another approval. As robots run, the system records task assignments, live observations, operator interventions, and site-state changes into one shared timeline that safety and operations teams can review in real time. When the incident stabilizes, it turns the robot data into an auditable clearance packet and after-action replay that a refinery can use for restart approval, insurer reviews, and future drill design. Over time, the platform becomes the source of truth for which robot workflow worked at which site condition, shrinking both exposure risk and downtime.

What's different. Incumbent robot OEMs sell chassis, autonomy modules, or single-vendor teleoperation consoles, but they do not become the cross-fleet system of record for clearance decisions at hazardous industrial sites. Generic field-service and incident-management tools also miss the core workflow because they do not model autonomy envelopes, operator overrides, or robot-generated evidence tied to human re-entry gates. This startup wins by owning the supervised mission playbooks, the evidence format a site uses to reopen safely, and the dataset of robot interventions and outcomes across dangerous-ground workflows.

Jobs to be done

flowchart LR
  Buyer[Refinery response lead] --> Pain[Unsafe slow post-incident assessment]
  Pain --> Product[Supervised robot ops and clearance layer]
  Product --> Outcome[Faster auditable crew re-entry]

Executive takeaways

• The strongest wedge is not robot hardware but the supervised workflow and evidence layer required before people re-enter hazardous refinery areas.
• Buyer urgency is real because the same sites already face confined-space, process-safety, and chemical-accident documentation burdens.
• Competition exists, but it is fragmented between OEM-specific robot stacks, vendor-neutral autonomy platforms, and industrial data systems rather than a purpose-built clearance OS.
• The right entry motion is evidence-first and supervisor-in-the-loop, not full autonomy replacing refinery safety authority.

Market definition

The relevant category is supervised hazardous-zone robot operations software for refineries and adjacent gas-processing plants: software that coordinates mixed mobile robots, enforces autonomy boundaries, captures operator interventions, and turns robot observations into auditable re-entry evidence for safety and operations teams.

Customer and buyer

Daily users are emergency-response leaders, operations-integrity teams, and robot-program owners at refineries. The economic buyer is typically the refinery manager, emergency response director, or VP of operations integrity who owns restart speed, worker exposure, and compliance risk.

Buying triggers

• A fire, gas release, confined-space event, or near miss forces the site to prove acceptable entry conditions and document who authorized re-entry. [9][11][14][16]
• Once robots are already used for inspection, the next bottleneck becomes cross-fleet coordination and evidence handoff rather than locomotion alone. [22][25][28][31]
• Critical-infrastructure operators increasingly face scrutiny on documented safety and preparedness, which favors systems that create durable audit trails. [13][19][39]

Willingness to pay

Buyers can justify meaningful spend if the platform prevents even a small number of hazardous manual entries or shortens post-incident assessment. Caltrans reports saving about $60,000 on one confined-space robot deployment, while bp and Woodside both frame robot inspection as a way to remove people from dangerous environments and speed decision-making. In refineries, that makes a six-figure annual software budget plausible even before counting a major incident avoided. [16][27][28][33]

Category dynamics

Growth signal 8.7% CAGR

Validation signals

• Shifters' seed round explicitly funds supervised autonomous ground robotic systems and one-operator mission coordination for dangerous environments.
• bp publicly says it would rather send a properly instrumented robot into hazardous environments than place people nearby.
• Woodside already uses Spot at Pluto LNG for hazardous inspections and regulatory visual inspection work.
• ANYbotics shows certified hazardous-area robotic inspection demand in oil and gas and gas-processing facilities.
• Caltrans demonstrates that confined-space robot use can save money immediately even before counting avoided injury or shutdown risk.

Regulatory & technical constraints

• Permit-required confined-space rules require acceptable entry conditions, attendant roles, and formal entry authorization.
• Process safety management rules require prevention of catastrophic releases and documented operating procedures around hazardous chemicals.
• EPA RMP obligations require accident history, worst-case release analysis, and emergency-response planning for covered facilities.
• Robots entering explosive atmospheres must meet IECEx or equivalent hazardous-area certification requirements.

The field splits into four camps. OEM ecosystems such as Boston Dynamics and ANYbotics tie software closely to their own robots. Vendor-neutral autonomy vendors such as Korial and Field AI aim to sit above hardware, but are broader industrial autonomy platforms rather than clearance-workflow systems. Critical-infrastructure platforms such as Gecko and Cognite centralize asset data, but do not own refinery re-entry approval logic. The opening is a compliance-first operations OS that turns mixed-robot activity into a trusted clearance record.

Why incumbents do not win by default

• Robot OEM stacks. OEMs win when the customer standardizes on one robot, but their orchestration and evidence tooling is optimized for their own hardware rather than mixed-fleet safety workflows.
• Vendor-neutral autonomy platforms. Korial and Field AI are the closest software adjacencies, yet both position around broad industrial autonomy and inspection rather than a refinery-specific clearance packet for human re-entry.
• Critical-infrastructure analytics platforms. Gecko and Cognite can become systems of record for asset condition and industrial data, but they stop short of owning supervisor approvals, autonomy envelopes, and incident-by-incident re-entry evidence.
• Incident and alarm software. Formant and adjacent incident tools can help triage physical operations, but they are not purpose-built for hazardous-zone robot missions or Ex-rated site procedures.

Hot Zone Robot Ops sells a supervised operations and evidence layer for refinery emergency-response teams that already use or are procuring ground robots. The immediate buyer pain is not robot locomotion; it is coordinating mixed robots during gas leaks, post-fire sweeps, and confined-space alarms while preserving human signoff and restart documentation. The first beachhead is North American refineries with 2-10 robots and acute pressure to reduce manual hot-zone entry. The product starts as a drill and post-incident clearance system because safety qualification and insurer trust are the main adoption bottlenecks, and the available research does not yet prove buyers will accept a new live command-path product on day one. TAM, SAM, and SOM are modest but venture-plausible only if the company becomes the workflow system of record in refineries and then expands into gas processing, LNG, and adjacent critical infrastructure. The strongest competitive position is vendor-neutral clearance workflow depth: OEM software is tied to one fleet, while industrial data platforms do not own re-entry approval logic. The largest disconfirming risk is that most early robot programs remain single-vendor and bundled, which would compress standalone software ACV and force distribution through OEMs or integrators. Another open gap is the lack of direct customer evidence on multi-vendor prevalence and budget ownership, so the first 90 days must test bundling, workflow priority, and willingness to pay before scaling headcount. Given the strong pain signal but unresolved packaging risk, this is a diligence-worthy industrial software opportunity rather than a high-conviction underwriting case today.

Problem

• Refinery safety and operations teams still stitch together manual entry, OEM consoles, contractor inspections, and spreadsheet logs when deciding whether crews can re-enter a hazardous area.
• Existing tools do not combine mixed-robot tasking, human approval gates, and auditable clearance evidence into one workflow that operations, safety, and insurers can trust.

Solution

• Provide a vendor-neutral command and evidence layer that coordinates the top refinery robot platforms through supervised mission playbooks for gas-leak perimeter sweeps, post-fire clearance, confined-space scouting, and valve-isolation verification.
• Convert robot telemetry, operator interventions, and site observations into a shared timeline, clearance packet, and after-action replay that fit existing refinery safety authority instead of replacing it.

Why we win

• The company is targeting the workflow incumbents skip: cross-fleet supervised clearance before human re-entry, where regulatory documentation and restart speed matter more than robot novelty.
• If it wins early sites, it can compound proprietary playbooks, integration templates, and clearance outcome benchmarks faster than OEMs or generic industrial data vendors.

Milestones

flowchart LR
  Wedge[Refinery drill and clearance wedge] --> MVP[Evidence-first multi-robot MVP]
  MVP --> Proof[Accepted clearance packets and faster remote assessment]
  Proof --> Expansion[Multi-site refinery expansion then gas-processing entry]

Founding team

Experiment roadmap

Risk assessment

What must be true

• At least three of the first five target refineries report real pain from mixed-vendor or mixed-payload robot workflows that OEM tools do not solve.
• A refinery buyer will approve a standalone or clearly itemized software budget of at least $200k ACV after a successful pilot.
• Safety teams will accept the platform's clearance packet in a formal drill or post-incident review without requiring a contractor to recreate the documentation manually.
• Supporting the first two or three robot platforms covers more than 70% of near-term target deployments.
• Drill, readiness, and routine hazardous-entry usage are frequent enough to prove ROI inside 12 months even if major incidents are rare.

Open diligence questions

• Who owns the budget line for this workflow today: refinery operations, emergency response, reliability, or the robot OEM bundle?
• How many current target accounts already run more than one robot, payload, or console in the same hazardous workflow?
• Which first workflow converts fastest to paid production: gas-leak perimeter sweep, post-fire clearance, confined-space scouting, or valve-isolation verification?
• What evidence format would a refinery manager or insurer accept as part of re-entry approval today?
• Which OEMs or integrators are willing to resell a neutral workflow layer instead of blocking it?

Model sanity

• Revenue engine. Base-case revenue comes from converting two paid design partners into 5 production sites by Q4Y2 and reaching 8 sites at a $360K blended ACV by Q4Y3.
• Must go right. The company must prove accepted clearance packets and preserve a standalone software line item before OEM bundling compresses ACV.
• Model breaks if. If sales cycles stretch to 12 months and Y3 ACV falls toward $300K, the downside case turns cash negative before the next raise.
• Next-round proof. A credible seed story is exiting Q4Y2 with 5 production sites, one partner-sourced deployment, and evidence that drills convert into annual licenses.

Scenarios

Sensitivity

flowchart LR
  Prospects --> PaidDesignPartners
  PaidDesignPartners --> ProductionSites
  ProductionSites --> SubscriptionRevenue
  ProductionSites --> UsageFees
  SubscriptionRevenue --> GrossProfit
  UsageFees --> GrossProfit
  GrossProfit --> OperatingCash

Flags: The model still exits Y3 slightly EBITDA-negative, so a seed round likely happens before self-funding. · ACV depends on buyers preserving a standalone software budget instead of forcing OEM or integrator bundling. · The market is niche: even the base case only reaches the researched 8-site refinery beachhead by Q4Y3. · Churn is heuristic rather than observed because the base case has too few live sites to establish retention history.

• Slow safety qualification. Refineries may require long validation cycles before a new robot-operations layer is trusted inside emergency procedures. Mitigation: Start with drills, after-action review, and insurer-facing evidence workflows that fit existing SOPs before taking over live command paths.
• Hardware fragmentation. Ground robots, payloads, and operator consoles vary widely, which can make vendor-neutral orchestration expensive to deliver. Mitigation: Constrain the first wedge to the top crawler and UGV setups used in refinery inspection and build adapter templates around those workflows.
• Irregular incident frequency. Some sites may struggle to justify software spend if they evaluate ROI only on rare emergency events. Mitigation: Price around drills, readiness audits, and routine hazardous-entry checks so the product creates recurring value between major incidents.