Business case · HeatSafe Voice Copilot

The case for HeatSafe

A voice copilot for people who need an answer while their hands are already full, and a business model tested on paper before anyone spends a dirham.

The problem

The information a site supervisor needs already exists. It is in the company SOP, the inspection checklist, the manufacturer's manual, a weather app, a regulator's website and a bulletin somebody emailed in March. All of it correct. None of it reachable while standing at the base of a scaffold in the sun.

The problem was never a lack of information. It is finding the right information, from the right authority, at the exact point of work.

Before a supervisor can say yes or no to a job, they check today's conditions, find the company rule, work out whether a regulation or an equipment limit also applies, compare the numbers, then explain the outcome to the crew. That is five steps and three documents, standing up, under time pressure.

HeatSafe removes the retrieval and the interpretation. It does not remove the human being who is accountable for the call.

What the research supports, and what it does not

PlanGrid and FMI found construction teams spending more than five hours a week searching for project information. Autodesk and FMI put bad data behind roughly $88.7 billion of global rework in 2020. Both are useful evidence that retrieval friction is expensive. Neither is proof of what HeatSafe recovers.

HeatSafe targets one narrow slice of that: the time spent finding and interpreting operational and safety information at the point of work. Everything past that gets measured in a pilot, not claimed in a deck.

What it answers, what it refuses

In scope

  • Voice questions over your loaded SOPs
  • Live wind, heat and weather
  • Regulation and manufacturer limits
  • Should we be doing this job now?
  • How do I do this job?
  • The source, named, every time

Deliberately out

  • Any autonomous ruling on safety
  • Stopping work by itself
  • Answering when no approved source covers it
  • Guessing when the audio is unclear
  • Inventing a number it cannot cite
  • Replacing the accountable supervisor

A safety copilot that bluffs once is worth less than no copilot at all. Refusing well is the product.

Who buys it

The budget sits with the Head of HSE, the Safety Director, the Operations Director or the COO. The people who actually use it are supervisors and frontline crew, which means adoption is won on site, not in the boardroom, and the two audiences need different proof.

The pricing hypothesis

Per worker pricing with a minimum site commitment. Per worker alone makes contracts too small to serve. Per site alone stops growing the moment the workforce does.

TierCommercial assumption
Paid pilotAED 5,000 per site per month. Up to 100 workers, one site, three SOPs, two use cases, 60 days.
StandardAED 35 per worker per month, minimum 250 workers per site. Minimum contract AED 8,750 a month, annual after the pilot.
EnterpriseAED 25 to 30 per worker per month at 1,000+ workers. Multi-site, SSO, APIs, audit logs, language packs.
OnboardingAED 15,000 to 50,000 once, depending on SOP volume, languages, integrations and security requirements.

One contract, run through

A 500 worker site at AED 35 a worker. The arithmetic below is contract maths on a fictional client, not a sales forecast.

17.5KAED monthly revenue
210KAED annual revenue
3.0KAED monthly run cost
83%illustrative gross margin
Voice, live context, weather, hosting. Before support, sales, onboarding labour and everything else a real company pays for.

The customer's side of it

If HeatSafe eventually saves ten minutes per worker per day across 500 workers and 22 working days, that is 1,833 hours a month. At a blended AED 30 an hour, roughly AED 55,000 of labour time against a AED 17,500 subscription, about 3.1 times over.

Every number in that paragraph is an assumption: the ten minutes, the rate, the adoption. It is a sensitivity scenario, not a promise. The honest version of this slide says illustrative, and then goes and runs the pilot.

The offer

Give HeatSafe one site for sixty days. Load three SOPs, pick the two use cases, and measure time to a verified answer, how many questions get resolved from approved sources, how many get escalated, and how often it correctly refuses. Those are the numbers that decide whether this becomes a company.

See the pitch deck ↗ See the build Back to HeatSafe

Written at the Dubai AI Hub Builder Lab #3, August 2026, alongside the build. Meridian Construction is a fictional demo client. Pricing is a hackathon hypothesis pending UAE customer discovery. Questions or ideas: say hi.

← Back to ankitabiswas.com