5 Questions Every CTO Should Ask Before Hiring a Prototyping Agency
You need a prototype. Your stakeholders want something they can click through before approving budget for the real build. You're evaluating vendors.
Every agency will tell you they're fast, flexible, and uniquely qualified. What they won't volunteer is the part where their "working prototype" is actually a Figma mockup, where timeline estimates assume zero scope changes, and where you don't own anything useful at the end.
These five questions cut through it. Pay attention to which answers you get versus what you asked for.
1. What's the actual timeline from kickoff to working code?
What they typically say: "We can get you a prototype in 4-6 weeks, depending on scope and your availability for feedback sessions."
The buried qualifier: your availability for feedback sessions. Every delay is now, at least partially, your fault. The real timeline, once you account for discovery meetings, wireframe reviews, developer handoffs, and QA cycles, is closer to 8 weeks. The Capgemini and Deloitte data backs this up — enterprise prototype engagements average 6-8 weeks from signed contract to deployed demo.
What an AI-driven approach delivers: Working code in 48 hours. The pipeline runs overnight: describe the business challenge, the AI generates a complete specification (database schema, feature list, architecture, estimates), scope negotiation is automated, and the prototype is built and deployed. No waiting for a design team, no waiting for a developer to become available, no review cycles before you can see anything.
What to listen for: Any qualifier that creates a dependency on your availability — "pending your feedback," "subject to scope confirmation," "assuming stakeholder access within 48 hours" — is a sign the clock doesn't start until they decide it does.
2. What do I own at the end — wireframes or deployable code?
What they typically say: "You'll receive the full Figma design files plus a working prototype, and all assets transfer to you at handoff."
"Full Figma design files" is not code. "A working prototype" from most agencies means a static HTML page with no backend — links that go nowhere, forms that submit to nothing, dashboards with hardcoded numbers. It looks like a working app. It doesn't behave like one.
Actual deployable code means real database tables, real API endpoints, forms that write to a real database, dashboards that query live data. The difference matters the moment a stakeholder tries to interact with it.
What an AI-driven approach delivers: Every ProtoForge-generated prototype is deployed at a real URL with a real PostgreSQL database. Each table in the specification gets a CRUD interface connected to live data. You can create records, read them, update them, delete them. It's not a simulation.
What to listen for: Ask explicitly: "If I wanted to deploy this to production today, what would that take?" If the answer involves more than a week of engineering work to convert the prototype to working code, what you're getting is not a prototype — it's a specification document with nicer visuals.
3. How much will scope changes cost me?
What they typically say: "We work on a fixed-scope contract, but any changes outside the defined specification are billed at our change order rate — typically $X per hour."
Translation: the initial estimate is for the version of the project that exists in the first week. Every insight you gain from seeing the prototype — every "actually, we need this to work differently" — is a billable change order.
This isn't malicious. It's how consulting economics work. The team priced the work against a fixed scope. Changing scope mid-build means re-doing work they already did. The incentive is to minimize changes, not maximize how much you learn.
What an AI-driven approach delivers: Because the spec is generated before the build starts, changes to the specification are iteration, not scope creep. Regenerate the spec. Rebuild the prototype. The cost is a subscription, not a change order.
What to listen for: Ask for a sample change order from a previous project. Agencies that are confident in their process won't hesitate. Agencies that bury change orders in fine print will give you vague assurances about "working collaboratively" instead.
4. Can I see a working demo before committing $50K+?
What they typically say: "We'd be happy to show you work we've done for similar clients, subject to any NDAs they've signed."
You will see a curated portfolio of their best work, selected by them, with context provided by them. This answers the question "can this team produce something good" — it doesn't answer "will they produce something good for this specific project at this specific timeline."
Try before you commit is the right standard. Any tool or agency that won't let you validate the core value proposition before signing a large contract is asking you to take significant risk without corresponding data.
What ProtoForge delivers: You can generate a prototype of your actual use case right now, before any commitment beyond a $29/month subscription. Submit your business challenge. Get a working prototype in 48 hours. See what you're buying before you decide whether to buy it.
This is the practical answer to question 4. If you're evaluating whether a prototyping approach will work for your specific problem, validate it on your specific problem — not on someone else's healthcare dashboard that vaguely resembles what you're building.
Try ProtoForge with your actual use case →
5. What happens after the prototype — handoff docs or a black box?
What they typically say: "We provide comprehensive handoff documentation including architecture decisions, component library, and development notes to ensure continuity for your internal team."
In practice: a PDF with screenshots, a Figma file your engineers have to reverse-engineer, and a 30-minute knowledge transfer call. Your internal developers will spend two weeks understanding what decisions were made and why before they can extend anything.
The black box problem compounds when you want to evolve the prototype into a real product. If the agency used their preferred stack, their preferred patterns, and their preferred shortcuts, none of which were documented or explained, your team inherits a foreign codebase with no institutional knowledge.
What an AI-driven approach delivers: The specification is the documentation. Every database table, every feature, every architectural decision is captured in a structured spec before a line of code is written. The prototype is generated from that spec, not the other way around. When your team extends the prototype toward production, they're working from a documented foundation — not reverse-engineering a codebase from a vendor who's already moved on to their next client.
What to listen for: Ask to see an example handoff document from a completed project. Redact the client name if needed. The quality of that document is the ceiling on what your team inherits.
The Honest Summary
These five questions surface the same underlying tension: traditional prototyping agencies are optimized for agencies, not for buyers. Fixed scope protects margins. Figma deliverables minimize build risk. Change orders monetize scope ambiguity. Vague handoffs create re-engagement opportunities.
None of this is malicious. It's just how the incentives are structured.
The AI-driven alternative isn't perfect either. You trade customization for speed. The generated prototype might not have every interaction polished, and complex integrations (Stripe, SSO, third-party data sources) are stubbed rather than live.
What you get instead: working code in 48 hours, at $29, with a real database, that you can extend toward production without starting over.
If you're evaluating whether to spend $50K on a prototyping engagement, spend $29 first.
Submit your use case to ProtoForge →
ProtoForge generates working software prototypes from business descriptions in 48 hours. Real database, real endpoints, real URL — not a Figma mockup. $29/month, no usage limits.