Architecture and scalability
Will what's been built cope with the next customer, and the next order of magnitude? Coupling, single points of failure, multi-tenancy and what breaks first under load.
The deck tells you what the startup wants to sell. The engineering tells you what it can deliver.
An independent assessment of a startup's product and engineering across ten dimensions, carried out by someone who has led more than 50 technical due diligences at CESAR, one of Brazil's leading innovation centres, for investment funds and large companies. For the investor who needs to know what they're buying, and for the founder who'd rather find out before someone else does.
Will what's been built cope with the next customer, and the next order of magnitude? Coupling, single points of failure, multi-tenancy and what breaks first under load.
What needs rebuilding, with estimated time and cost. Kept apart from what's merely ugly and can live on for another two years without harm.
Data model, quality of what's collected, cost per run and how dependent it is on a model provider that could change its price or terms.
Real exposure: attack surface, secrets management, access control, personal data processed without a lawful basis. Not box-ticking.
Cost per customer and its curve, predictability of the bill, vendor lock-in and how much of it is reversible.
Continuous integration, testing, observability, delivery cadence. Teams without these ship fast at first and stall halfway.
Who keeps the product running if one person leaves tomorrow. Concentration of knowledge, real seniority and dependence on contractors.
What's been built versus what's being sold. The gap between the two is the commercial debt nobody records.
Licences incompatible with commercial use, code written by contractors without assignment of rights, abandoned dependencies and contracts that follow the product.
Whether what's been promised to the market fits the team that exists, in the time stated. It almost never does: the question is by how much.
Every finding with its severity, business impact and estimated cost to fix. Written for whoever makes the investment decision, not just for the people who write code.
The summary for the committee: what's a blocker, what's a concern and what's acceptable for the company's stage.
A live discussion of the findings, with room for the questions the report didn't anticipate.
An assessment that arrives with a delivery proposal attached has an incentive to find problems. This one ends when the report has been delivered and discussed. If it then makes sense to commission delivery, that's a separate contract, negotiated from scratch, and you're free to hire anyone, including someone else.
Without this the timeline doesn't hold, and a due diligence based only on what's in the pitch isn't worth what it costs.
Code, history and issues. The history often says more than the code.
A technical conversation, not a sales pitch.
Actual cost, not an estimate. It's where the architecture gives itself away.
Ideally before first access, by all parties.
Tell me which startup it is, what stage it's at and what decision depends on this assessment. I'll reply with available dates and the access I need.