Well media kit
Well is the new interface to run your company through conversation. The story, the product, the founders and the assets, all cleared for press use.
Overview
What Well is and why it exists, in a form you can quote.
Well, the conversational interface to run your company
Bank, invoices, customer relationships, taxes, admin work: Well brings every tool a company uses into one conversational interface, reachable from Claude, ChatGPT, WhatsApp, the browser or the web app. Well connects those tools, turns them into one context graph, and acts inside them with the founder's approval at every step.
Well was founded in 2025 by Maxime Champoux and Bastien Blanc. It offers a new way to run a company: instead of switching from one piece of software to the next, the founder asks a question or hands over a task in a conversation. Well covers seven domains of running a company, brought together in one context graph, and acts directly in the connected tools rather than only talking about them.
A rare founding pair. Maxime Champoux and Bastien Blanc met at iBanFirst and have lived digital banking for more than ten years. Between them they have built four core banking systems, at Qonto and Fintecture among others. With Well, they apply that experience to the whole of running a company: connect every tool, keep its memory, and act inside it.
In practice. Well answers a complaint founders who use AI agents make today: most agents reason well but cannot act in real systems. Well exposes its connectors through an MCP server usable from Claude, Cursor or Codex, a browser extension and a web app, all built on a graph that resolves every counterparty, document, transaction and conversation into one entity per company.
Boundaries first. Business tools usually centralise all the data first, then carve out access rights. Well does the opposite: each company the founder runs starts in its own space, with its own data, connectors and memory. What Well learns about one company is not passed to another, and the founder decides what to bring together in one view.
The founder asks a question or hands over a task in Claude, in the browser or in the app, and Well acts on their real tools, across seven domains:
How much cash do we have today, and how many months of runway? Well reads the real balances of every connected account, computes the monthly burn and projects cash.
Categorise transactions, chase missing receipts, match payments to invoices, prepare the month-end close: the hours founders describe as the black hole of their week.
Find missing supplier invoices in inboxes and portals, show the bills due this week, prepare a customer invoice ready to send.
Prepare the data for tax filings from books that are already reconciled, instead of retyping everything at the deadline.
Who owes us money, and since when? Well ranks receivables by age, finds the customers who weigh the most, and links each payment received to its invoice and to the contact in the CRM.
Link payslips, contracts and time spent to the rest of the company, to know what each team and each project really costs.
Everything Well learns about the founder and the company stays in memory: nobody explains their context twice, from one tool or one conversation to the next.
“We are entering a new world: the boundary of the tools used to run a company is being redrawn. The agent becomes omniscient about your business. All its knowledge is resolved, put in context and centralised, and it is driven from a new conversational interface. Your bank account, your invoices, your customer relationships, your taxes, and above all the admin work that eats a founder's week: everything is handled in one place, through conversation. It is a unique opportunity for European application companies to reinvent how a business is run, straight from Claude, ChatGPT or WhatsApp.”
Fact sheet
The essentials, one line each.
- Name
- Well
- Founded
- 2025, Paris
- Headquarters
- Station F, Paris
- Founders
- Maxime Champoux (CEO) and Bastien Blanc (CTO)
- Team
- 12 people (September 2026)
- Product
- A conversational interface to run a company, across seven domains: FP&A, bookkeeping, invoicing, taxes, CRM and revenue, time tracking and payroll (coming soon), and a personal and business brain
- Surfaces
- Claude, ChatGPT, WhatsApp, a browser extension, the web app and an MCP server
- Markets
- France first, then the United States, then the rest of the EU
- Pricing
- Usage-based: a monthly plan with an AI budget in tokens, AI billed at cost. Three plans, from 9 € or 12 $ a month excluding tax
- Website
- wellapp.ai
Describing Well
The sentence, the short forms and the boilerplate. Copy them as they are.
The sentence
Well is the new interface to run your company through conversation: it connects every tool the company uses, brings them into one context graph that keeps their memory, and acts inside them with the founder's approval.
Short forms
Boilerplate
Short
For the foot of an article or a release
Well is the new interface to run your company through conversation. FP&A, bookkeeping, invoicing, taxes, CRM and revenue, time tracking and payroll (coming soon): Well connects a company's tools, brings them into one context graph that keeps their memory, and acts inside them with the founder's approval, from Claude, ChatGPT, WhatsApp, the browser or the web app. Founded in 2025 in Paris, at Station F.
Long
For a backgrounder or a profile
Well is the new interface to run your company through conversation. A founder of a small company works across ten to fifteen apps, and none of them can answer the question that matters on a Friday night. Well connects those tools, then does the hard part: it resolves the same supplier across the payment tool, the bank and the ledger into one counterparty, and attaches what was said in a meeting to the company it concerns. On that graph, Well answers questions and acts: it reconciles, categorises, chases missing receipts and prepares the month-end close, with the founder's approval at every step and every figure traceable to its source. Each company starts in its own workspace, and the founder decides what to bring together. Well was founded in 2025 by Maxime Champoux and Bastien Blanc, who have built four core banking systems between them, and is based at Station F in Paris.
Vocabulary
- Conversational interface
- Where the founder runs the company: a conversation, in the tools they already use, rather than one more app to open.
- Context graph
- The single, resolved picture of a company that Well builds from its tools: every counterparty, document, transaction and conversation, linked.
- Business brain
- What Well remembers about the founder and the company, so nobody has to explain their context twice.
- Boundaries first
- Each company starts in its own workspace. Bringing several together is the founder's explicit choice.
- Founder in the loop
- Every action Well takes goes through the founder's approval.
Vision
Three convictions, and the principle that holds them together.
The two founders met at iBanFirst. Maxime Champoux then spent seven years at Qonto, joining around employee 30 and staying through a team of 1,600, where he built the core banking system and structured a product organisation of about 150 people. Bastien Blanc became VP Engineering at Fintecture, where he built the company's core banking system. Well comes from a conviction they share: an agent only reasons as well as the context it is given, and running a company already means doing by hand, every day, what is now called context engineering.
The line Well is built on
We are the average of the people we meet. Our agents are the average of the context you feed them.
An old line about people, applied to agents.
Three convictions
Connect every tool the founder uses
A founder should never have to choose between good tools and one agent that understands all of them. Well connects the bank, invoicing, payments, the CRM and the books in one place, without the founder explaining their company to each tool again.
“Your tools never talked to each other. At Well they finally do, and you approve what they do together.”
Concentrate knowledge to speed up intelligence
Scattered knowledge is useless while it stays scattered. From those tools, Well builds a living, factored context graph that keeps the why behind every decision, so an agent reasons with the company's real memory, not with fragments.
“Well links your cash, your books, your invoices, your tax filings, your customers, your payroll and everything you know about your company into one living graph, so the agent never has to ask you again what you already told it.”
Invent the interface of tomorrow
The interface of tomorrow is not one more app to open. It is the context already there, in the tools the founder uses every day, Claude, Chrome, the web, acting on their behalf and waiting for their approval before each action.
“No more switching screens or explaining everything again: the agent acts where you already work, and you approve every action.”
The principle behind all three: boundaries first
Business tools index the whole organisation into one graph, then filter what each person may see: the boundary is a permission added afterwards. At Well, the boundary is the starting point. A founder who runs three companies has three workspaces, each with its own brain. Every record belongs to one workspace from the moment it is created, and the server enforces it. Bringing several companies into a group view is an explicit choice by the founder, not a permission setting. An agent works in one workspace at a time, and only in those the founder has authorised.
“At Well, every company starts in its own space, and you decide what to bring together.”
Product
Seven domains of running a company, six building blocks, and every surface Well reaches.
Well captures a company's data (bank, invoices, email, documents, CRM, payroll), enriches it and resolves it into one context graph, then acts directly in the connected tools: reconcile a transaction, chase an invoice, prepare a close. The founder approves each step.
Six building blocks
Agentic harness
The layer that connects the AI model to the company's tools and guarantees their coverage.
Context and data model
Adopted in seconds and shaped to each company: a modular model that adapts to how it works.
Surfaces
Usable wherever people already talk and work together: Claude, ChatGPT, WhatsApp, the browser, the web app.
Next generation UI
One interface to drive every tool, with deliberate design choices.
Workflows and skills
Assemble the building blocks that keep the company running.
Agents
Delegate at scale to expert agents, under the founder's control.
Surfaces
- MCP server
- api.wellapp.ai/v1/mcp, OAuth 2.0 with dynamic client registration. Usable from Claude, ChatGPT, Cursor and Codex, and listed in Claude's connector directory.
- Browser extension
- Reads the tools that have no API.
- Web app
- app.wellapp.ai
The founder stays in the loop. Every action the agent takes goes through an approval, not full autonomy: Well's direct answer to the unease founders express about agents that act unsupervised.
Key figures
What runs on Well today, and the size of the problem it solves.
On Well today
700,000+
records in total
200,000+
company records (customers, suppliers)
200,000+
invoices
80,000+
bank transactions
80,000+
documents
70,000+
contacts
Records on the platform in September 2026, net of the team's internal, test and demo workspaces. Well does not publish its number of workspaces or users, nor any internal financial figure.
The product
~11,400
banks in the aggregated network
55
business objects in the context graph, grouped in 8 data families
7
domains of running a company
The market
The problem
Owners spend about seven hours a week on bookkeeping and more than twenty hours a month on financial tasks. Admin takes roughly a third of the work week, and small companies run 70 to 150 SaaS apps depending on the study. The founder is the integration layer.
SCORE; Xero; JumpCloud 2025; Zylo 2025
Founders
Bios and portraits, cleared for press use. Both founders give interviews in French and English.
Maxime Champoux and Bastien Blanc are a rare pair: they have built digital banking for more than ten years, and between them they have built four core banking systems, two each. They met at iBanFirst, which is where the working relationship starts. Core banking is the hardest system in financial software to get right, and almost nobody has built one twice.
Maxime Champoux
Co-founder and CEO
Maxime Champoux is co-founder and CEO of Well. He spent seven years at Qonto, joining at about 30 people and helping scale it to 1,600, where he built the core banking system and ran the integration program.
Maxime Champoux is co-founder and CEO of Well. He started in payments at iBanFirst, then spent seven years at Qonto, joining around employee 30 and helping scale the company to 1,600 people. There he built the core banking system, structured a product organisation of about 150, and ran the integration program that connected Qonto to the tools its customers already used. That decade taught him where a business keeps its truth, and where it loses it. He founded Well in 2025 with Bastien Blanc to give every founder one conversation above all of their tools.
Bastien Blanc
Co-founder and CTO
Bastien Blanc is co-founder and CTO of Well. He was VP Engineering at Fintecture, where he built its core banking system, after a decade in data infrastructure.
Bastien Blanc is co-founder and CTO of Well. He began at iBanFirst, where he met Maxime Champoux, then became VP Engineering at Fintecture and built its core banking system. A decade in data infrastructure shaped the conviction behind Well: connectivity without meaning only produces errors faster, so the data model comes first. He owns the resolution layer that turns a company's tools into one context graph.
Media library
Portraits, logos and product clips. Use the logo on a surface that gives it contrast, and never redraw the symbol.
Portraits
Black and white, free to use for press coverage with the credit © Well. Each download is the JPEG at full resolution, 1376 pixels wide.
Maxime Champoux and Bastien Blanc, office, meeting table
DownloadMaxime Champoux and Bastien Blanc, office rooftop
DownloadMaxime Champoux and Bastien Blanc, interview setting
DownloadMaxime Champoux, office lounge
DownloadMaxime Champoux, office rooftop
DownloadMaxime Champoux, interview setting
DownloadMaxime Champoux, interview setting, close
DownloadBastien Blanc, office
DownloadBastien Blanc, office rooftop
DownloadBastien Blanc, interview setting
DownloadThe office
Black and white, free to use for press coverage with the credit © Well. Each download is the JPEG at full resolution.
The office under its glass roof
DownloadThe office kitchen
DownloadThe office rooftop, facing Sacré-Cœur
DownloadThe office building
DownloadThe office courtyard
DownloadLogo
Product clips
Screen recordings of the product, cleared for use as b-roll. Scroll sideways for the whole set. Each one also runs on its changelog entry.
Enrich records from your tables
7.3 MBSelect rows in any table and press Enrich to fill missing data.
Connect your bank accounts
2.8 MBAbout 11,400 banks, with transactions synced in real time.
Expose your business stack over MCP
1.6 MBAn MCP server that lets AI tools read and write live business data.
Filtering across workflows and tables
2.3 MBFilter documents at the router level and narrow records in tables.
Brand kit
well-brand-kit.zip
Every logo, symbol and app icon in SVG, PNG and JPG at the common sizes, plus a favicon set.
FAQ
The questions journalists ask first.
- What is Well?
- A new interface to run your company through conversation. Well covers FP&A, bookkeeping, invoicing, taxes, CRM and revenue, time tracking and payroll (coming soon), and keeps the company's whole context in memory. It acts directly in the connected tools, with the founder's approval.
- Who founded Well, and when?
- Maxime Champoux and Bastien Blanc, in 2025, in Paris.
- Where is the data kept, and how is it protected?
- Well starts from the boundary. Each company the founder runs has its own workspace, with its own data, connectors and memory. Every record belongs to one workspace from the moment it is created, and the server enforces it, never the user or the agent. The founder chooses which workspaces to bring together in a group view. An agent works in one workspace at a time, and only in those the founder has authorised. What Well learns about one company is not passed to another. Well does not train AI models on its customers' data.
- How many people work at Well?
- 12 people, including the two founders (September 2026).
- Is Well profitable? What is its revenue?
- Well does not disclose its revenue or profitability.
- Where is Well based?
- Station F, Paris.
- Is Well available outside France?
- Well launches in France first, then in the United States, then in the rest of the EU.
- How is Well priced?
- Well is usage-based. Each plan combines a monthly subscription with an AI budget measured in tokens, the unit AI models bill by. Each task handed to Well uses a measured share of that budget, shared by the whole team in the workspace. AI is billed at the model provider's cost, with no margin. Connecting and syncing tools costs nothing, and MCP connectors are unlimited. Three plans, excluding tax: Starter at 9 € or 12 $ a month with 3 million tokens, Business at 29 € or 35 $ with 10 million, and Scale at 69 € or 80 $ with 25 million, sold through the sales team. Starter and Business have no commitment, monthly billing and a 14-day free trial.
- How much data runs on Well?
- More than 700,000 records on the platform (companies, invoices, contacts, transactions, documents), net of internal and test workspaces. Well does not publish its number of workspaces or users.
Contact
One person for the press, who answers directly.
Press
Maxime Champoux
Co-founder and CEO
We answer within one business day.
Interviews in French and English.
Op-eds
Long-form pieces by the founders, available to publish.
Seven pieces by the founders, available to publish. Write to maxime@wellapp.ai to receive one.
The category piece
The Fourth Interface: Why You Will Uninstall Your Apps
The interface got unbundled from the app. Reads three eras of interface change from fifteen years inside banking, then names the incentive flip that makes this era different.
The moat piece
No Ontology, No Love: The Data Model Is the New Moat, and You Should Own Yours
Enterprise AI fails not from lack of ambition but from never formalising the data model, and a business should extend the ontology it already owns rather than outsource it.
The velocity piece
The Double Diamond Is Dead. Long Live Design Engineering.
The classic design process broke once shipping velocity outran design review. The design engineer's job moved from front-loaded prototyping to standards curation and tooled anomaly review.
The method piece
Writing Culture Is Back. But Not For Us.
The return of documentation culture is not about company scale, it is about audience. The reader is an agent now, and moving from a wiki to markdown is what that shift looks like in practice.
The why-now piece
The SaaSpocalypse Wasn't About AI. It Was About Architecture.
The 2026 repricing was an architecture gap, not an AI event. The companies that lost the most value had bolted a conversation box onto database queries instead of building AI-native systems.
The customer story
AI That Remembers Your Business: How Chat Memory Changes Everything
The retention problem in AI tools is the cost of re-explaining yourself. Opens on a real customer anecdote, a Munich owner correcting the same fiscal-year assumption over and over.
The counterweight
Vibe Coding Is a Shein Haul: Why Speed Without Craft Is a Landfill
Speed only compounds when the builder understands the system. Otherwise it produces landfill-grade debt.