You are Priya Sandhu, 33, product co-founder of Tessellate Labs, a three-person start-up in Toronto. It is Tuesday 16 June 2026 and your advisory board has just frozen money you were counting on. This week you have to decide what the team spends its first weeks on.
The situation
Tessellate's only asset is MosaicQA, a machine-vision model that detects micro-defects in extruded aluminium and polymer profiles from cheap cameras mounted above a production line. Your technical co-founder, Émile Doucet, built it in eleven weeks with AI-assisted coding tools. You have a working prototype, two unpaid pilot conversations and no customers.
There is C$420,000 in the bank and you spend C$30,000 a month: about fourteen months. The board's memo, which arrived overnight, has one sentence underlined: "Optionality is now a liability." It will not release the remaining C$90,000 of a federal grant until it sees a single commercialization plan, backed by evidence, on 14 September. Six months of exploration have left four possible markets: aluminium extruders, automotive-parts quality assurance, food-packaging inspection and a "defect API" for any manufacturer. None has been tested with a buyer.
Of some twenty beliefs the company rests on (who has the pain, how much it hurts, what they would pay, who signs the cheque, whether the model survives plant dust at line speed), only three have been tested with a real buyer. Émile wants to spend C$120,000, four months of cash, turning MosaicQA into a finished product. Every week spent building rather than selling uses about C$7,000. Meanwhile a US competitor has raised US$4 million to sell general-purpose visual inspection.