AI-native marketing

A first campaign founders can approve with confidence.

Ventora turns product signals into a ready-to-review campaign, explains its judgment, and stays legible once the campaign is live.

The product decision

Replace a long configuration exercise with a campaign receipt: one plan, clear trade-offs, and approval only where human judgment matters.

The same moment, before and after

Drag to compare the real chat-led setup against the plan Ventora now arrives with.

Ventora campaign review with a prepared plan, AI recommendations, and readiness checklist
Earlier Ventora marketing flow with a chat panel and mostly empty campaign settings
BeforeAfter

The interaction decisions

Four product calls that make campaign setup faster, clearer, and easier to trust.

Founder interviewsCompetitor teardown — 6 AI ad toolsSupport-ticket themesSaaS onboarding benchmarks
Decision 01

Start with a plan.

Before — Chat asked for goal, budget, targeting, and creative.
Issue — Founders stalled before seeing the campaign.
Now — Ventora reads the site first and drafts the plan.
Decision 02

Show the why.

Before — AI recommendations looked like final answers.
Issue — People distrust advice they can’t question.
Now — Every pick shows confidence and reasoning.
Decision 03

One receipt, not tabs.

Before — Campaign setup lived across separate panels.
Issue — Too much setup before the decision felt real.
Now — Goal, budget, audience, and creative sit together.
Decision 04

Keep launch visible.

Before — Live campaigns disappeared into a dashboard.
Issue — Invisible automation feels like losing control.
Now — Progress, timeline, and next steps stay visible.
Plan

A point of view, ready to review

Product and landing-page signals become a complete campaign proposal before the founder has to configure anything.

AI Design Approach

Production-ready today. Agentic by design.

Traditional UI
Manual configuration
AI assistance
Drafts on request
Today
AI collaboration
Draft, explain, approve
Next
Agentic workflow
Proactive operating loop

How it was built

Human in the loop
Review before publish
Explainability
Why this recommendation?
Recommendations
Budget and audience
Progressive disclosure
Advanced controls on demand
Regeneration
Alternative campaign drafts
Editable AI
Every draft stays mutable
Next: strategy, creative and risk agents converging on one shared decision — with approval still human.

Final solution

Interactive prototype

Open

AI Lab · Fintech concept study

Viber Pay, designed from context rather than a prompt.

I explored how a messenger could make sending and receiving money feel native to the conversation, without turning trust into a chatbot problem.

Open the live prototype

01

Conversation-native

Transfers start from what people are already saying, not an isolated finance surface.

02

Receive-first

Gifting and claiming create a useful reason to enter the wallet before asking for setup.

03

Inspectable trust

Scam protection and AI suggestions show their reasoning at moments with real consequences.

Interactive prototype

Explore the payment moments in context.

Live GitHub Pages build

How the work was directed

The work stayed legible at every handoff.

context/
market.md
philippines.md
social-rituals.md
gift-culture.md
trust.md
payments.md
viber-strengths.md
ai-principles.md

System context

Viber strengths

Messaging graph · group chats · existing identity · low friction

Avoid

AI everywhere · chatbot · financial advice

Prioritize

Invisible AI · context-aware · conversation-native

The outcome: a working concept with 10 product interactions and a documented decision trail.

Evidence-led
AI lab / landing

A B2B landing built from one brief.

An iGaming aggregation concept, generated end to end and then edited down.

Open landing

01

Stack over catalogue

The page explains what an operator plugs into, instead of leading with a game count.

02

Archetypes, not demographics

Player segments are framed by how people actually play, so the offer reads as product rather than inventory.

03

Built in one pass

Structure, copy and layout were generated together, then edited — a test of how far a single brief carries.