A human in the loop where it matters.

AI agents can move software projects from idea to MVP faster than ever, but there’s a major difference between vibe coding and agentic development: vibe coding treats the AI agent as both architect and developer, producing quick results without your input, which can result in a rats nest of incoherent code under the hood. That works for a proof of concept, but it becomes dangerous when the MVP touches real customers, business processes, important CRMs, payment systems, and fragile hosting environments that need long-term support. Agentic development, on the other hand, means inserting an architect into the generation process every step of the way: they define the system; break work into small, comprehensible tasks; review the output; catch bad assumptions; protect maintainability; and make sure the agent’s code aligns with the larger business processes. In this way, I use AI as a force multiplier inside a human-led development process: fast where speed helps, careful where risk matters.

You want to build a legacy of code that will withstand unpredictable change, not change with it unpredictably. Don’t just let the agent build it: let an expert direct the agent.

Generative AI helps us pressure-test for ideas that resonate.

Iterative branding uses generative AI to explore and prototype visual identity. We can start with a handful of complementary ideas (whether they’re emotions we’re trying to evoke in our audience, cultural motifs we want to incorporate, or artistic styles we want to adopt) to produce a traditional moodboard that outlines our creative vision, and then use AI tooling to rapidly generate visual directions. Iterative branding through genAI allows us to quickly hone in on an “Ah ha!” moment that our visual designer can run with. All the while, the process of experimentation is driven by human judgment and intent. This approach is especially valuable for early-stage brands, creative projects in flux, or teams navigating a rebrand without the budget or risk tolerance for expensive (and therefore irreversible) bad decisions.

Rather than replacing visual designers and the process of art direction, generative AI helps us pressure test for ideas that resonate, long before the work hardens into real world assets and design guidelines.

Ecosystem design is not about adopting more AI for the sake of it; it’s about adopting the right amount in the right places.

Ecosystem design focuses on how generative AI can fit into a broader creative or technical environment within your business. Instead of members of your team adopting isolated tools to yield sporadic outputs across the organization, I can help you design workflows that connect only the most relevant tools to your data, and put the right people in the pilot’s seat of those tools, so that a single, coherent system can emerge that accelerates your overall business processes. This includes auditing your workflow to uncover acceleration opportunities, then selecting appropriate third party applications (or configuring local pipelines) and defining how these tools interact with existing design or engineering processes.

Ecosystem design is not about adopting more AI for the sake of it; it’s about adopting the right amount in the right places, so the system tells a coherent story about how work gets done.

Every ecosystem is different. This work benefits teams who feel overwhelmed by the pace of AI development and don’t have the enterprise budget (or the technical know-how) to take advantage of what the cutting edge has to offer.

A genuine collaborator that can extend your reach.

Generative AI can dramatically accelerate content production when it’s guided by clear creative intent. But the power doesn’t come with the button pushing; anyone can do that. Instead, content production at scale requires careful prompt engineering and a deep understanding of the character of generative models (some LLMs write more creatively than others; some image models are better at photorealism than illustration) to ensure outputs meet a standard of quality that won’t come across as inauthentic to audiences on the receiving end. The selection of fine tunes to bias with aesthetic preferences on the image side, or the construction of RAG pipelines to prime LLMs with the right datasets, are both part of the technical setup involved in establishing a production workflow.

There are many other considerations that go beyond that technical setup, including the question of the longevity of third party integrations in a field where the technology is constantly changing, and an ever-evolving legal landscape.

But once it’s in place, genAI goes from being a thoughtless automatic content engine to a genuine collaborator that can extend your reach.

There’s room to educate, and so it’s important to get ahead of the debate.

The attitudes toward genAI in the public sphere are as diverse as they are fierce. According to worldwide polls as of 2026, meta analysis shows that your average consumer is skeptical of, or at least indifferent to the technology, despite vocal evangelists and even more vocal anti-AI activists. This means there’s room to educate, and so it’s important to get ahead of the debate.

Policy development addresses the structural questions surrounding generative AI adoption head on: what should be used, how, by whom, and under what constraints. I help organizations articulate internal guidelines that balance experimentation with accountability, covering issues like data privacy, ownership, disclosure, and ethical use. These policies are tailored to the organization’s values rather than copied from generic templates.

Drafting a policy such as what I’m describing is a first step before a formal legal review and organization-wide adoption. It’s especially valuable for teams whose lack of clear guidelines can lead to inconsistent use, or worse: outright conflict. By framing policy as a living document—open to revision as tools evolve—organizations can engage generative AI thoughtfully, without freezing innovation or ignoring responsibility.