Shambhavi Singh

Designer · Builder · Researcher

Clarity before canvas.

I design at the intersection of concept and system — and then I build the thing.

VisiWise Optimization

Current work

Digital operations,
built for business.

Founder & CEO · Phoenix, AZ · since July 2026

I lead end-to-end design — concept, wireframe, prototype, final delivery — for client brand and web identities, grown to three active B2B clients.

0
active B2B clients
0
profile views in 6 months
delivery capacity, no new headcount

Visual identity and communication systems designed end to end in Figma and Adobe Creative Cloud, translating client goals into on-brand digital and print-ready assets.

Underneath that sits structured content and information design — semantic architecture and content strategy — which took one client's Google Business Profile to 3,210 profile views and 483 search appearances over six months.

And I documented the studio's design and communication workflows well enough to train a volunteer operator on them, doubling delivery capacity without adding headcount.

visiwiseoptimization.com ↗
visiwise · client dashboard
Active B2B clients
3
Brand + web identity, retained
Shipped end to end
2
Extension · live multi-user room
Delivery pipeline
ConceptWireframePrototypeDelivery
Figma · Adobe Creative Cloud
GBP profile views
0
6 months, one client
Search appearances
0
Same window
What moved it
Semantic architecture and content strategy — structure first, copy second, listings last.
One system, two interfaces
Administrator and audience views over a single source of truth. Events below are sample traffic.
adminpinned the market open notice
guest_04joined the room
guest_11asked about the 4pm close
adminpushed a data refresh
Delivery capacity
Workflows documented, one volunteer operator trained — no added headcount.
Operators trained
1
On design + comms workflow
Client growth
0→3
Since July 2026
Multi-Agent Epistemic Research System

Truth hastwo sides.

Two AI agents independently research opposing sides of any question using live web sources, structured belief graphs, and semantic contradiction detection. Not another chatbot — a belief accumulation engine.

0
research phases
0
live sources per debate
2
agents, zero shared context

The interface surfaces calibrated uncertainty instead of an answer. Node weight is confidence; edges are the claims each agent used to get there; the whole thing streams over SSE as the research runs.

Where both agents reach a claim neither can settle, it stays marked contested — you see the disagreement rather than an average of it.

Designed and built end to end. Next.js, React, GPT-4o, Exa Neural Search.

belief graph · streaming
FOR AGN contested YOU claim

The Research

Two original concepts

A sole-authored position paper, peer-reviewed and accepted to NordiCHI ’26, proposing two original frameworks for decentering AI in human–nature–technology design.

AI does not merely imitate nature — it structurally inherits it. Intelligence whose architecture reproduces the recursive organizational principles found from dendrite to network to ecosystem. A metaphor borrows an image; an architecture inherits a mechanism.

When attention and capital concentrate on one technology at once, it becomes culturally load-bearing — and load-bearing things are misusable. At sufficient scale they can override commons-based rights that were never up for negotiation.

Accepted August 2026 · NordiCHI ’26, the 14th Nordic Conference on Human-Computer Interaction · ACM Adjunct Proceedings, forthcoming October 2026. The paper is not yet published — ask Silva for a summary, or get in touch.

Selected Work

Hover to read

Let’s make something real.

Open to full-time roles, freelance, and collaboration.