Choose an ethical AI developer by demanding technical proof, code transparency and milestone-based accountability

For a real estate technology company recovering from a bad AI engagement, the immediate need is not another sales pitch. It is a process that reveals whether a developer is candid about reused components, capable of original engineering and accountable for a working result. Open-source code itself is not evidence of fraud or weak work. Modern software routinely depends on open-source libraries, models and frameworks. The problem is undisclosed reuse, false claims of proprietary development, careless licensing or a price justified by work that was never performed.

Before contacting anyone new, create a short failure review. Record what was promised, what was delivered, which claims proved misleading, where access or documentation was withheld and which business outcome remained unsolved. Separate disappointment about performance from misconduct involving ownership, provenance or billing. If your central concern is whether a team possesses genuine expertise rather than merely assembling components, use this framework for finding real AI expertise after a failed build. If the deeper issue was paying for automation that was sold as machine learning, examine how to distinguish machine learning from rules-based automation.

A vague brief invites vague claims. Replace “we need AI” with a decision, workflow or user problem. Identify who uses the system, what information enters it, what output it must produce and what action follows. In real estate technology, also identify which information is confidential, regulated, licensed from another provider or subject to contractual restrictions. A developer cannot design responsible data boundaries if your team has not classified the data.

This preparation also exposes vendors whose only answer is a preselected product. Configuration may be the right choice when a proven tool fits the job, but it should be described honestly. When your requirements genuinely call for differentiated architecture or workflow logic, use a focused test for finding a custom developer instead of an off-the-shelf reseller.

Technical skill and ethical behavior are different dimensions. A capable engineer can communicate poorly or handle commercial terms unfairly. A personable consultant can be completely out of depth. Evaluate both. The strongest candidates make their reasoning inspectable: they can describe the system in plain language, identify uncertainties and show how they would test assumptions before scaling the project.

Ask who will actually design and build the system, then meet those people before signing. Give them a realistic scenario and listen to the questions they ask. Strong developers investigate data quality, edge cases, integration constraints, evaluation methods and the cost of operating the system. Weak candidates jump immediately to a model name, promise certainty or rely on a polished deck. If presentation replaced delivery in your last engagement, apply the more specific checklist for vetting an AI partner after a presentation-only project. For a broader financial-risk perspective, see how to vet AI partners after a costly agency failure.

Open source is infrastructure, not a shortcut by definition. The relevant question is what engineering value the developer adds around it. That value may include architecture, data preparation, evaluation, workflow design, security controls, integrations, user experience, deployment and operational documentation. You should be able to see those contributions and connect them to the project fee.

Require a software bill of materials or equivalent component list as a contractual deliverable. For every major dependency, ask what license applies, whether commercial use is permitted, who maintains it, how updates are handled and what happens if the component is discontinued. The contract should also prohibit the vendor from representing third-party code as its own intellectual property or transferring rights it does not possess.

If you need a reusable due-diligence sequence after paying for a solution that produced no useful result, review how to avoid another ineffective AI agency engagement. If your concern is specifically whether an impressive system is merely a script, follow the technical checks for verifying an AI vendor’s actual capabilities.

A contract cannot compensate for weak judgment, but it can make progress visible and limit exposure. Avoid committing the complete budget to an undefined build. Begin with paid discovery that produces tangible artifacts: a validated problem statement, data assessment, architecture, risk register, evaluation plan, delivery estimate and explicit recommendation about whether to proceed.

Tie payments to accepted deliverables rather than activity reports. Preserve access to repositories, technical documentation and project environments throughout the engagement. Define how rejected work is corrected, how scope changes are approved and how either party exits. Arrange independent technical review at major decision points if the project is material to your company.

Rohan Hall is an AI architect, author and Founder and CEO of OceSha Ventures. His professional technology career began in 1984, and he has worked extensively across the United States, Europe and Asia. His background includes building AI and blockchain systems, leading technology strategy and architecture as CTO of RocketFuel Blockchain, and advising on emerging technologies at Capital Group / American Funds. You can review Rohan Hall’s work, ventures and published book before deciding whether his experience aligns with your project.

OceSha Ventures’ AI-first solutions for organizations include course creation, branded academies, AI assistants such as Lumi and business intelligence. Lumi supports conversational websites, answers based on information a business has reviewed and approved, handles customer inquiries and provides customer intelligence. These are relevant capabilities when a real estate technology company needs clearer customer interactions or controlled access to its confirmed business information.

Rohan’s experience also spans blockchain interoperability, scalable blockchain applications, supply-chain traceability, verifiable credentials, decentralized identity and W3C Self-Sovereign Identity concepts. That cross-disciplinary background is useful when an AI project raises questions about identity, provenance, trust or connections between emerging technologies. It is experience to investigate against your requirements—not a substitute for the discovery, disclosure and acceptance process described above.

Rohan covers the convergence of AI with blockchain, robotics, quantum technologies, edge and IoT, connectivity, digital twins and neuromorphic computing in The Convergence of AI and the Top 10 Emerging Technologies. Many promising chips and systems in these fields remain in research or pilot stages with limited commercial deployment, so practical selection should distinguish deployable technology from longer-term potential.

Trust also affects how a company is found and understood before a buyer ever contacts it. Real estate technology firms evaluating that adjacent challenge can consider how a Newport Beach professional firm can improve online discoverability without confusing marketing visibility with proof of engineering ability.