Skip to content

Algolia case study

From UI factory to strategic partner

A shopping assistant built in Agent Studio: asked what to wear to a beach wedding in July, it suggests a linen maxi dress, a short-sleeve cotton shirt with tailored trousers, and a straw hat and sandals, with four related sources

Moving design upstream at Algolia, and taking Agent Studio from zero to one.

As Senior Manager, Product Design, I led design across Algolia’s search, AI and developer products.

Design came in too late to matter

1 : 3

product designers to PMs and engineering managers

AI was core to Algolia’s strategy, but its tools lacked a usability focus. Engineers saw design as UI support, and PMs said design added value too late to change outcomes.

One designer in every lane

Service desk
Efficient, but reactive, with little influence
Fully embedded
High influence, but not scalable at one designer to three
Swimlanes
Coverage and cohesion, if the design leads are strong

We chose swimlanes: one designer per major product domain, owning it end to end with its PM and engineering managers, and a Principal bridging the lanes for coherence.

I led the six designers, from mid-level to Principal, promoted one from Senior to Staff, and introduced critique, progression frameworks and AI hack days.

Agents in weeks, not months

Agent Studio’s landing page in the Algolia dashboard: “Build and publish AI agents with ease”, bring your own LLM, and a Create your first agent button

I led design for Agent Studio, a framework for building AI agents on Algolia’s search. Developers set up an agent’s purpose, tone, data and tools in the dashboard, then ran it with one API call through their own LLM provider. It launched in public beta in August 2025.

Agent Studio’s agent preview: the Shopping assistant template, its key capabilities, and a live chat to try it before using it

Across the whole product surface

Ask AI
Turned the docs search bar into an assistant that answers
Algolia MCP Server
Let agents like ChatGPT and Claude query Algolia’s APIs
Agentic Components
InstantSearch, reimagined as a foundation for chat assistants
Visual merchandising
No-code tools for arranging search and category pages
NeuralSearch
The vector and keyword search engine underneath it all

Agent Studio was the key launch, not the only one. With design across the whole product, my team worked on every release between November 2024 and May 2026.

The agentic search bar on a demo outdoor store: typing “Running shoes” shows suggestions, products, and Ask AI questions such as “What are the best running shoes for trail running?”

Three rocks that defined great work

The Rules editor before: “It all starts here” over a grid of trigger cards, then “What do you want to do?” over a grid of strategiesThe Rules editor after: a plain list of triggers under “Trigger”, then the strategies under “Strategy”

Before

After

We set three Big Rocks for the whole team. The first, usability, simplified screens like the Rules editor, and took design system pattern reuse to about 70%.

  • Usability: complex AI tools that feel effortless
  • Technical depth: designers fluent in APIs and agents
  • Cohesion: trust between humans and agents, by design

What it added up to

$6M+

Of ARR influenced by Agent Studio

4.7/5

Customer satisfaction score for Agent Studio

100k

Agent Studio API calls a week

0 → 5

Product domains with design in roadmap talks

What I learned along the way

Gloss black walkie-talkie with a translucent lilac antenna

Design maturity isn’t about adding process, it’s about shifting how people see its value.

If I did it again, I’d start earlier on storytelling. Narrative shifts perception faster than structure.

Next project

Travel Communities

Set the vision and led the team through it. 1M monthly active users in under a year, and 12-⁠week retention doubled.

View case study →
White universal travel adapter with lime pins
Book an intro call