Skip to content
Provolu, home
Let’s talk ↗ Client portal Français

Work

Solutions built around the problem.

There isn’t always an off-the-shelf product for a specific problem. We start from the process you already have, find what slows the work down, and use technology where it adds real value.

Some of this work was done for organizations that remain confidential. We describe their needs, our approach and the work delivered without naming them. Dre Katty and MyQuotaBag are presented by name. Other organizations remain confidential.

Featured

02 / Studio product · Website · iOS app

Explore the project

MyQuotaBag

MyQuotaBag · Website in motion

A world of its own.
Down to the smallest detail.

A Parisian watercolor world connecting a website and an app for collectors.

MyQuotaBag is independent and is not affiliated with Hermès.

01 / Website · Art direction

Explore the project

Dre Katty

Dre Katty · Website in motion

Trust starts
before the first visit.

A distinctive identity, clear service pages and a natural path to getting in touch.

03Data · Sales intelligence · Dashboard

Sales intelligence

The opportunities hidden in data you already have.

Illustrative mockup: the accounts and data shown are fictional.
The problem
A large history of sales data existed, but it was hard to use. Trends, gaps and opportunities stayed buried in it.
The approach
Import, clean and structure the existing data, then build a tool that quickly shows trends, gaps, changes in behaviour and opportunities that call for action. The goal isn’t more charts. It’s less noise, and attention pointed at what can actually shape a decision.
What it brings
  • import of existing data
  • cleaning and structuring
  • trend analysis
  • customer-to-customer comparisons
  • detection of meaningful gaps
  • opportunity spotting
  • results put in context by a human eye
  • an action-oriented dashboard

In shortFrom a file that’s hard to work with to a view that helps you decide where to look, and where to act.

Scale

More than180,000

records: the dataset the architecture was tested on.

More work

04

Cross-team product collaboration. Years of scattered information, turned into one shared view.

Product operations · Collaboration · Management

The problem
A team was tracking dozens of important items, requests, priorities, decisions, across spreadsheets, documents, emails and meetings. Everyone in their own way. Nobody saw the whole picture.
The approach
A shared structure first, then a lightweight web tool to display, filter and share it. The tool is built to be used during working discussions: the essentials on screen, the details within reach without cluttering the view.
What it brings
  • requests gathered in one place, each with a permanent identifier
  • priority and effort tracking
  • decisions and their history
  • links back to the sources
  • a view designed for meetings
  • a common language between teams with different trades

In shortLess ambiguity, less lost information, and better-structured discussions across teams.

Scale

Nearly100

complex follow-up items, brought into one shared, searchable structure.

05

Team operations follow-up. Simple, week after week.

Custom application · Workflow · User experience

The problem
Recurring tasks, responsibilities and approvals lived in scattered lists, messages and reminders. It was hard to see what was done, and what was left.
The approach
An application that brings it all into one simple weekly view: what’s complete, what remains, what needs attention. It was designed around how people actually work, rather than around the structure of generic software.
What it brings
  • centralized weekly follow-up
  • responsibilities and recurring tasks in one place
  • quick additions during the week
  • simple visual statuses
  • history and approvals

In shortFewer forgotten tasks, less manual follow-up, and a shared view of what needs to get done.

06

Intelligent document extraction. From document to data, in seconds.

Artificial intelligence · API · Automation

The problem
Every receipt or document had to be read, then retyped by hand, before it could be used in an application.
The approach
A process that analyzes the document, automatically finds the relevant information and turns it into structured data, ready to use. AI serves one precise task, where it removes most of the data entry, at a very low cost per analysis.
What it brings
  • automated document analysis
  • extraction of the information that matters
  • normalized data
  • integration into an existing workflow
  • targeted use of APIs
  • an architecture designed for a low cost per operation

In shortRepetitive data entry becomes an almost instant operation.

07

Technical control interface. A complex system, simple decisions.

Technical UX · Product · Information architecture

The problem
An interface had to present many parameters, states, alerts and controls, without passing that complexity on to the person using it.
The approach
Functional design work. Information is ranked by importance, details appear only when they become useful, and technical data becomes information that tells you the situation and the action to take.
What it brings
  • simplified technical data
  • a clear information hierarchy
  • alert and notification logic
  • details revealed at the right moment
  • fewer unnecessary actions
  • design based on real operating conditions

In shortComplex technology becomes more intuitive, faster to understand and easier to use every day.

08

An AI-orchestrated work environment. Shared memory, clear responsibilities.

AI operations · Agents · Knowledge management · Automation

The problem
Working with AI one conversation at a time means repeating the context, and getting decisions that contradict each other from one session to the next.
The approach
An architecture where several specialized agents contribute to the same projects from a shared context: decisions, project status, responsibilities, available access, actions under way. Each agent has a defined role. A regular log keeps track of what’s done, what’s left, and what needs a human.
What it brings
  • a central project registry
  • structured memory, independent of any specific model
  • specialized agents with defined responsibilities
  • access and permission rules
  • continuity from one session to the next
  • traceable decisions and sources
  • an action log
  • structured handoffs between agents

In shortLess context to repeat, fewer conflicting decisions. AI becomes a real way of working, rather than a series of disconnected conversations.

A similar problem? Let’s start with what you have.

If one of these problems looks like yours, write to us. The first step is always to understand the process that already exists.

Start a project

or write to info@provolu.com