01 / AI imagingRetouchless
Consistent product imagery without giving up control.
The AI image editor that helps commerce teams recolor and retouch product visuals without giving up creative control.
retouchless.xyzOperator-led company studio / Grenoble, France
We buildcompanies from 0→1and stay to operate them.
FokaLabs turns real problems into software companies. We shape the product, build the technology, ship to real users and operate what comes next.
Product · Engineering · Operations
- 0
A real problem
Usually one we have met ourselves. Framed, scoped, worth a company.
Bring us a problem The next one starts here - 1
A first product in real hands
Shipped, used, paid for or forked. This is where most studios hand over.
- Retouchless
WNOHÉ
Clozup
START UGA
beloteXbelote
- n
Operated for the long run
Real users, real data, infrastructure, support and the next version. We run it and keep building it.
Helmai
Carbonable
SqareX
Baaton
notifyd
Maison Craie
Philoé Paris
ScopeRail
AIHEALTHCLIMATEDEV TOOLSCONSUMER
01 / How we build
Four steps.
The last one never ends.
The same loop runs behind every company here, from the first conversation to daily operations. Product, design and engineering move together, so a small operating team ships with the leverage of a much larger one.
- 01
Frame
Find a real problem worth turning into a company. Understand the user, the decision and the limits before writing code.
- 02
Build
Product, design and engineering move together. One team, one source of truth, and a first version scoped to be useful.
- 03
Ship
Get the first useful product into real hands. Real users, real data and real feedback replace assumptions.
- 04
Operate
Run it, automate it, learn from it, keep building. Infrastructure, support and the next version are part of the work.
1 → n
Most studios hand over at one. We stay for what comes next.
02 / In plain words
Not an agency.
Not a venture studio.
Three models that look alike from the outside. The difference is what happens after launch, and who carries it.
| Model | Agency | Venture studio | FokaLabs |
|---|---|---|---|
| Starts from | A client brief | A thesis and outside founders | A problem we have met ourselves |
| Builds | What was specified | A company to spin out | The product and the company around it |
| After launch | Hands over | Steps back once the company is independent | Keeps operating it |
| Team | Billable staff | Founders recruited per company | One operator with an AI-native stack |
Partner builds exist, ScopeRail and START UGA among them. They follow the same loop and the same standard.
Principle
The real product
starts after the demo.
AI-native tools let a small operating team build with the leverage of a much larger one. But shipping faster only matters if the product survives real users, real data and real operations. We build for that from day one.
Built AI-native03 / Selected work
Built for
real use.
01 / AI imaging
02 / Voice softwareWNOHÉ
Voice software built to survive real work.
The macOS app that turns voice into text, notes and tasks with local-first processing.
wnohe.com
03 / ClimateCarbonable
Climate contributions with a record people can audit.
The platform companies use to fund, track and audit their climate contributions.
carbonable.io04 / Companies & products
Built here. Still running.
Thirteen products and companies, in order of impact. Each row carries a status you can check today: live, operating, or used by other projects.
- For
- Dental practices, clinicians and front-desk teams
- Problem
- Missed calls, emails and day-to-day disruptions arrive through separate channels. Context gets lost before the team can decide what happens next.
- Outcome
- One system captures each signal, links it to the patient, proposes the next action and leaves the final decision with the practice.
- For
- E-commerce, studio and portrait teams
- Problem
- Colorways and cleanup create repetitive production work, while generic AI makes art direction hard to control.
- Outcome
- Produce consistent image variants faster while keeping the source, direction and final choice in human hands.
- For
- Professionals who write, meet and operate all day
- Problem
- Capturing notes and turning speech into useful text interrupts the work it is meant to support.
- Outcome
- Turn voice into structured text, tasks and notes while keeping sensitive work local-first.
- For
- Companies and climate-project operators
- Problem
- Climate contributions are difficult to monitor, reconcile and explain with a trustworthy record.
- Outcome
- Manage, track and report every contribution through one auditable system.
- For
- B2B sales and growth teams
- Problem
- Account research and useful personalization take time, while generic automation damages trust.
- Outcome
- Turn research into contextual outreach that a team can inspect and control.
- For
- Dental groups and multi-site practice operators
- Problem
- Stock, purchasing, incidents and compliance live across disconnected tools and local habits.
- Outcome
- Run every site from one operational view with shared processes and traceable actions.
- For
- Software teams operating with AI agents
- Problem
- Agents and humans lose context when work is split between chat, tickets and tools built for only one side.
- Outcome
- Coordinate both through one programmable source of truth.
- For
- Developers and autonomous software agents
- Problem
- Background work needs dependable notifications without another heavy platform or channel-specific integration.
- Outcome
- Send and query operational notifications through one small, programmable service.
- For
- UGA students, associations and campus teams
- Problem
- Events, initiatives and associations are scattered across channels that students do not consistently find.
- Outcome
- Give campus life one searchable, accessible place to be discovered.
- For
- People and designers looking for singular interior pieces
- Problem
- Made-to-measure textile work is hard to specify, visualize and source outside opaque bespoke processes.
- Outcome
- Turn a personal brief into a traceable, handmade piece from a French workshop.
- For
- People looking for a clearer scalp-care routine
- Problem
- Scalp concerns are often treated with fragmented products and little continuity.
- Outcome
- Make diagnosis, product choice and routine feel like one coherent care path.
- For
- Belote players and groups of friends
- Problem
- Online card rooms often add friction around a game that should feel immediate and social.
- Outcome
- Get a table running quickly and keep the experience focused on playing together.
- For
- Product teams embedding AI into complex, multi-tenant software
- Problem
- A generic copilot can read the docs. It cannot see the current screen, live business state or the access boundaries that decide whether an answer is right for this user.
- Outcome
- Combine hybrid retrieval, read-only live tools and deterministic access controls to return precise, source-backed answers within each user’s authorized scope.
05 / The operator
One operator.
A studio's scope.
Ramzi Laieb founded FokaLabs after co-founding Carbonable, which the studio still holds a stake in. Product, engineering and operations for every company here, with an AI-native stack doing the work of a much larger team.
I build and operate software companies from 0→1: product, engineering and operations.
Founder / FokaLabs
What one operator covers
- 01
Product
Framing, design, roadmap and pricing.
- 02
Engineering
Architecture, code, infrastructure and security.
- 03
Operations
Support, data, growth and the next version.
French Tech Rise
French Tech Alpes regional laureate, then a national finalist in the deeptech edition held at the French Ministry of Economy.
Singapore FinTech Festival
Carbonable selected for Bpifrance’s French fintech delegation.
Paris-Dauphine
Invited by the FinTech Chair to explain how software can make environmental action verifiable.
EY Start-up of the Year
Regional finalist with Carbonable in Auvergne-Rhône-Alpes.
06 / Field notes
Notes from
the work.
What we have learned building and running products across AI, operations and software infrastructure.
View all field notesFrame an AI product before choosing a model
A practical method to define the user, decision, boundaries and success signals of an AI product before discussing models.
Read the guide →Orchestrate data, rules, models and people
How to design an operable AI workflow by separating layers, placing human arbitration and preserving a useful decision trail.
Read the guide →Automate repetitive work without hiding exceptions
A product architecture for automating the stable path, exposing exceptions and building a usable improvement loop.
Read the guide →Software sovereignty: keep the data, keys and choice
A concrete framework for building a portable AI product that is local-first where it matters and able to change providers without losing its core.
Read the guide →
07 / FAQ
The short
version.
An operator-led company studio. We build software companies from 0→1 and stay to operate them, across AI, healthcare, climate, developer tools and everyday life.
An agency starts from a client brief and hands over. A venture studio launches outside founders and steps back once the company is independent. FokaLabs starts from problems we know, builds the product and the company, and keeps operating them after launch.
Most products start inside the studio. We sometimes build with a partner when the problem is clear and the work can become a real product. ScopeRail, built for Sextan, and START UGA started that way.
Yes, but we take on few partnerships. We look for a clear problem, a committed operator and a useful role for FokaLabs. The contact form is the right place to start.
Yes. We open-source infrastructure when it can help beyond one product. notifyd, our notification server, runs in six projects today.
08 / Contact
Start with a
real problem.
Tell us what you are building, where it is stuck and why it matters. A few clear lines are enough.
Direct route[email protected] ↗