A working capability.
A defined workflow or product experience in an agreed environment. Clear boundaries around users, data, integrations, and what the prototype is intended to demonstrate.
03 / A working prototype
Prototype Lab
Your external skunkworks for a consequential AI product bet. We build a working prototype that lets you examine the capability, test the assumptions, and decide what deserves further investment.
A build with a question to answer
Make the next bet
possible to examine.
You have a product opportunity that matters, but the core team is committed elsewhere—or the idea needs room to develop before it enters the roadmap.
Avery takes responsibility for a contained prototype: product definition, the agreed build, and its evaluation. You provide a sponsor, domain access, and someone who can accept the result.
We start with the uncertainty the prototype must resolve. A useful result may support investment, expose a constraint, or show that an existing product is the better choice.
What you receive
An evaluation you can inspect.
The agreed test cases, observed results, limitations, and operating assumptions. Enough to challenge the demonstration and decide what to investigate next.
A path beyond the demonstration.
Architecture, dependencies, setup instructions, and the agreed code or licensed access. What would be required for further development, including remaining engineering work.
The implementation follows the problem
Existing technology.
New work.
The useful combination.
Where it fits, we can adapt or integrate technology from our own lab, including Humfrid’s company-context capabilities. We can also build a new prototype around your workflow and the tools you already use. A Humfrid license is not a requirement.
We establish what can be demonstrated today and what needs to be built. Existing components, third-party services, and custom code are identified in the scope.
The result can sit under your brand, including within an agency’s client engagement. White labeling describes the commercial arrangement; the work still begins with a specific user need and acceptance criteria.
Inspect our current technologyFrom question to evidence
A contained bet.
An explicit finish.
Frame the question
Bring the workflow, its intended user, what you need to learn, and the decision date. Include a budget range and the engineering capacity or access already available.
Agree the experiment
Define the prototype, test cases, acceptance criteria, environment, and handover. We confirm feasibility, a project price or staged quote, and a delivery date before work begins. Scope changes require agreement.
Build, evaluate, and hand over
Review the work at agreed checkpoints. At the end, inspect the prototype and findings together. Further development, production rollout, and ongoing support are separate commitments.
A good fit
- A consequential product or workflow question with a sponsor.
- A contained build that can resolve a real uncertainty.
- Access to representative, approved data or a useful synthetic substitute.
Choose a different engagement when…
- You need an independent recommendation from existing evidence: Decision Factory.
- You need an ongoing owner inside your existing team: Embedded Delivery.
- You need full production operation or a service-level guarantee: discuss a separate scope before committing.
Before the build
The practical questions.
Can you build something unrelated to Humfrid?
Yes, where the problem fits our applied AI and product expertise. We assess the proposed workflow and technical requirements first. Humfrid is one possible starting point, not a required platform.
Who owns the code and the result?
We agree rights before work starts: ownership or licensing of custom work, Avery’s existing technology, third-party components, and the rights needed for handover and further development. White labeling alone does not transfer underlying IP.
Can the work carry our brand?
Yes, by agreement. We establish branding, client contact, required approvals, confidentiality, IP rights, and support responsibilities in the scope.
How much does it cost, and how long does it take?
We quote against the uncertainty, build, and evaluation required. Share the intended result, budget range, and decision date; we return a proposed scope, price, and timing after confirming feasibility. The quote identifies any staged commitments and external service costs.
Is the result ready for production?
The standard result is an evaluated prototype in the agreed environment. Security hardening, scale, production integration, monitoring, and ongoing support require their own scope. The handover identifies those remaining needs.
Must we buy Decision Factory first?
No. Start directly with Prototype Lab when the immediate need is a working proof. A decision review or ongoing delivery engagement is only added if it has a separate purpose and scope.
Licensing & strategic fit
When the opportunity
goes further.
Some conversations start with a capability we have already built. We are open to technology licensing, integration, and confidential discussions about an asset acquisition, acquihire, or broader strategic combination where there is a strong fit.
Start with the capability and the strategic reason for the conversation. A paid prototype or decision review is not a prerequisite.
Discuss a strategic fitWhat do you need to prove?
Tell us the workflow, the uncertainty, and when the result needs to inform a decision.