THE CHALLENGE
In the last two years, we’ve worked on products where AI isn’t a feature, it’s the foundation.
From financial research platforms to land valuation tools and intelligent matchmaking assistants, we’ve helped shape products powered by large language models, automation, and real-time data. That shift changed how we design, not just the interfaces, but how we think through complexity, structure, and behavior.
Designing for AI calls for a different mindset. One where intelligence is a core material, something to shape, structure, and translate into real value for real users. It’s not about optimizing known patterns, it’s about defining new ones from scratch.
We partnered with Brightwave, Datalab, Novita and Endeavor to bring this new kind of product to life. Each had its own context. But the core challenge was the same: How do you make something as complex and dynamic as AI feel clear, usable, and trustworthy?
HELPING ANALYSTS ASK BETTER QUESTIONS AND GET BETTER ANSWERS
Brightwave turns data rooms and SEC filings into structured insight. The system reads like an analyst, surfaces what matters, and backs it with source.
We joined at concept stage to help define a product where the interaction model reflects how analysts think, layered, iterative, and skeptical. Designing meant matching the logic of investment workflows with the constraints of LLM reasoning.
What we shipped
Mapped how analysts build conviction, looping through questions, reframing, and evidence stacks
Prototyped multi-turn interactions to stress-test memory, logic retention, and conversational depth
Built a modular UI designed to grow with the model's capabilities without breaking usability
fig. 2 (Table view → Add documents flow)
fig. 3 (Synthesis)
fig. 4 (Synthesis)
We designed a research interface that holds up under pressure: fast, traceable, and built for the way analysts think.
fig. 5 (dashboard)
MATCHING FOUNDERS AND MENTORS WITH PRECISION
Endeavor Brain helps founders find the right mentor. Its AI assistant suggests matches based on expertise, context, and goals.
We helped improve the early version of the platform to feel like guidance, not guesswork.
What we shipped
Refined matching logic with better filters, richer inputs, and clearer prioritization
Designed flows that feel conversational and intentional, less form, more feel
Proposed backend improvements to stabilize recommendations and prep the system to scale
fig. 6 (Onboarding)
The result is an assistant that suggests with purpose and earns it.
fig. 7 (chat + feedback)
fig. 8 (Narrow search filters)
IN BRANDING WE TRUST
When AI is invisible, branding becomes the surface where people form trust.
We brought that principle into two distinct builds: Novita, made for developers, and Datalab, shaped for data teams. Different audiences, same demand for clarity. With Novita, a platform offering AI infrastructure, we created a brand system that spoke dev language: fast, technical, and clear.
fig. 9 (Novita website)
fig. 10 (Social media)
fig. 11 (wordmark)
With Datalab, where AI transforms unstructured docs into structured data, we went sharp and exact. Every element spacing, weight, color, designed to reflect precision and discipline.
fig. 12 (Datalab website)
In both, branding wasn’t decoration. It was infrastructure.
fig. 13 (symbol)
fig. 14 (digital mockup)
SMARTER ISN’T ENOUGH
Designing AI-native products means designing for evolving systems. That’s not just a UX challenge, it’s a product one. Interfaces have to carry logic, handle ambiguity, and still feel intuitive.
Across every project, our job was to turn intelligent behavior into clear interaction. If AI is the engine, design is the interface to belief. That’s how AI becomes usable. And how products become believable.
fig. 15 (dashboard)