Rebuilt a failing beta into a platform that influenced $6M in revenue. Set the vision that defined Algolia's agent strategy and SDK roadmap.
Scope
0→1 rebuild: architecture, UX patterns, and cross-org alignment
Co-designed API schema with ML engineers
Built AI-powered experiences & defined SDK component scope
Aligned ML, engineering, PM, legal, and business stakeholders
Impact
$6M influenced ARR
Established foundation for Algolia's agent product strategy
AI interaction model now adopted across Algolia Dashboard
Vision used in customer demos and shaped SDK priorities
Context
A private beta with low traction
Discovery
Three structural blockers, not surface problems
🔥
Backend driven UX → high friction to value
The product mirrored backend architecture, not user mental models. Users had to think like engineers to get value.
🔥
Lack of observability → low trust
Users couldn't see how agents behaved in production. No visibility meant no confidence to go live.
🔥
User flexibility vs. Subscription model
User needs (LLM choice, configuration control) directly conflicted with the subscription model, making scale economically unsustainable.
A big decision
The pivot decision: start from zero
Discovery made clear the blockers were structural. Incremental fixes wouldn't move the needle. Partnering with PM and engineering, we made the call to rebuild.
✅ Rebuild 0→1
8 weeks to MVP
Scalable foundation for future agent builder
Enable a monetization model tied directly to API usage
❌ Fix existing product
Product limited to RAG only
Revenue ceiling remains
Limited UX improvement opportunities tied to API design
Approach #1
Shifted the team from backend thinking
to user outcomes
The team was deeply technical but lacked a shared picture of who we were building for. I ran a storyboard workshop with PM, engineering, and ML - before we make aby architecture decisions.
Result
This shifted our approach from backend-driven structure (Config → Tools → Deploy) to outcome-driven design (Goal → Capabilities → Test).
Approach #2
User goal driven API design
Before any technical implementation, I designed the complete user journey and information architecture. This became the baseline for API schema design.
1
Defined new information architecture & user mental model
2
Co-created the API schema with engineers
3
Backend implementations + UX work in parallel
Clean backend-to-frontend mapping
This required upfront alignment across teams, but reduced long-term rework and prevented the UX from drifting back toward backend-driven abstractions.
Core platform features
Building a platform that can scale across products and teams
1️⃣ Configure and test side-by-side
Problem
The API-driven UX required users to configure prompts, data sources, and tools across separate pages, then assemble them into an agent at the end.
Solution
Introduced the Agent object as the central organizing unit. Users now configure instructions, tools, and memory in one place - and test the agent's response in real time alongside it. What was a multi-page assembly process became a single, continuous workflow.
Trade-off
Intentionally excluded advanced technical controls at launch to prioritize clarity for business users. Complexity introduced progressively based on real usage data.
2️⃣ Ability to monitor your agent and iterate
Problem
Lack of observability made it difficult to trust, debug and iterate on agent behavior.
Solution
Built-in monitoring dashboard showing key health metrics such as response time, search calls, and user engagement trends. Users can also review each conversations in case they need to investigate deeper.
3️⃣ Tighter ecosystem integration → direct revenue impact
Problem
Agent Studio was isolated from Algolia's core products, limiting both user value and monetization.
Solution
Integrated Agent Studio with Algolia's flagship products. Each agent now generates API usage that contributes directly to the revenue stream, turning the platform into a monetization driver, not just a product feature.
Post MVP launch
Adoption revealed where the platform needed to grow
Agent Studio had proven itself as a prototyping environment. The next phase was clear: evolve it into a production-grade platform, with the SDK, experience primitives, and control layer to match.
✨
Customers didn't know what "good" looked like
Enterprise customers needed a reference. A real picture of what an AI-powered ecom or media experience should look like before committing to a build.
🚨
Customers needed more control over how their agents think and behave
Monitoring gave visibility. But what customers needed next was control over agent behavior, guardrails, and cost. Without it, internal sign-off to go live remained out of reach.
Define the category
Built the vision that showed customers what to build, and how to build it
AI-powered ecom & media reference experiences
No reference existed for what a truly AI-native ecom or media experience should look like. Without one, customers built cautiously and sales conversations lacked a north star.
What this unlocked:
Became the primary artifact in customer demos and sales conversations
Defined which components the SDK team would build, the de facto SDK scope document
Aligned internal roadmap around a tangible vision, not a written spec
AI prototyping as a design decision tool
The most critical interaction to nail was the AI agent trigger inside a standard keyword search bar.
I tested two distinct logics in Google AI Studio to expedite the process.
Intent-based
Agent activates when query exceeds 3 words and autocomplete returns no strong match
Persistent
Agent entry point always visible within the autocomplete search box
Agent Studio evolved into a production-grade platform
In parallel, I designed the next evolution of Agent Studio, expanding it from a configuration tool into a platform that could support the full vision at production scale.
Experience builder
Customers compose and customize AI-powered front-end experiences directly within Agent Studio, connected to their agent configuration.
Controls & Guardrails
Rate limits, token controls, turn count caps. The layer that converts a prototype into something an enterprise team can sign off on shipping.
Measuring impact
From zero to a platform that defined the category
0M → 6M
Influenced Revenue
4.7/5
CSAT
100K+ / week
API calls
AI interaction model adopted across Algolia Dashboard
SDK scope and component priorities defined from vision prototype
Cost guardrails unblocked enterprise production adoption
Influenced 2025/26 AI product roadmap
Reflections
What I learned
✨ What worked
Co-designing APIs with engineers to mirror the user’s mental model was key to making the experience intuitive, scalable, and resilient.
Vision storyboarding helped shift the team from backend-driven abstractions to outcome-driven design, creating shared clarity before committing to architecture.
🪄 What I'd do differently
Quantitative discovery: We lacked baseline metrics. I would invest 2 weeks upfront in quantitative analysis & baselines settings before launching a new version.
Proactive research: Post-launch iteration was reactive in the beginning. Would implement continuous research earlier.