Forging,
writing,
thinking.
We share our methods, our client cases, what works and what we got wrong. No intrusive newsletter, no pop-up. Read what speaks to you, sort by topic.
All our reads.
34 articles, sorted from newest to oldest.
Your CRM is useless. It is not the software's fault
Fewer than four salespeople in ten really use it. The problem is not the tool, it is who the tool serves.
The cash sheet is right. It dates from Monday
A daily question, a weekly answer. And the red line for an assistant plugged into your accounts.
The technical bid you have already written
Your past answers are a corpus. The cost of a tender is not writing, it is finding the right paragraph.
Your clients have already told you. In writing
Reviews, emails, forms, tickets: the signal exists, it is simply spread out. How to read it together.
Your CRM is not wrong. It is empty
One week in four is not a discipline problem but a data-entry one. And a copilot that invents makes it worse.
GEO: getting cited by ChatGPT, Perplexity, Claude
After SEO, GEO. How your pages become answers in generative engines — without keyword stuffing 2.0.
RAG: feeding an assistant with your documents
Indexing, chunking, citations. The principles for an assistant that cites your sources rather than making them up.
AI Act: what applies to your company
Timeline, obligations by risk level, penalties. How to map your AI systems without panicking.
AI agents: the 2026 promise unpacked
Beyond the chatbot: what an agent actually does, where it still breaks, and the right scope to give it this year.
AI in sales: the winning bet
From quote to signature, where AI accelerates selling without distorting it. Four levers, two pitfalls.
AI in the HR function of mid-sized companies
Recruiting, labor law, payroll, training: what AI actually absorbs, where it remains fragile.
Multimodal in mid-sized firms: demo or production?
Photos, diagrams, scanned invoices, audio. What works in production today, what's still a demo.
Generative AI and cost reduction
The four areas where the drop is measurable, the two where the promise has almost never been kept.
Generative AI in 2026: where we stand
Beyond the hype: what works, what still doesn't, and the quiet shift of 2026.
Calculating and maximizing the ROI of an assistant
Honest method, typical ranges, classic estimation mistakes. From spreadsheet to bottom line in 18 months.
Measuring effectiveness: KPIs that matter
Five readable indicators, two to avoid, and the cadence at which to review them to steer.
How to succeed with your assistant project
Seven steps, from scoping to daily routine. The pitfalls that kill a well-launched project.
Integrating an assistant into your infrastructure
SSO, NAS shares, DMS, CRM, ERP. The right pivot points, the right formats.
10 questions to ask your AI vendor
Hosting, data, models, contract, reversibility. The questions that filter out the serious players.
Prompting like a pro
Five techniques to phrase questions that get the right answers, based on real examples.
Documentation makes the good assistant
An assistant is only as good as what it's given to read. The principles for preparing your corpus.
GDPR and AI assistants: keys to compliance
Article 28 processing, DPIA, data subject rights. The method to stay clean.
Why choose a sovereign AI
Cloud Act, sensitive data, jurisdiction. When sovereignty is required, and when it's optional.
AI sovereignty: what it really means
Hosting, models, jurisdiction, logs. Dismantling sovereign-washing in seven points.
Five concrete cases in public services
Civil registry, planning, finance, HR, communications: where AI truly helps, where it should still be avoided.
Deploying an AI assistant in a local authority
From political commitment to daily use: seven steps for a deployment that lasts.
An assistant for citizens
The promises, the pitfalls, and the right scope to honor the public service commitment.
City hall: an HR assistant for the HR department
Statutes, payroll, union law, career management. What AI absorbs in public HR.
City of 35,000: 20 years of council minutes
Story of the "municipal memory" project: OCR, indexing, political guardrails, intranet rollout.
E-invoicing: documenting smartly
PPF, PDP, Factur-X formats. How to prepare your firm to respond, without drowning the partners.
Documentation is also about relationships
Why your engagement reports become a strategic asset, not just a binder.
Turn your PDFs into a mission assistant
From scattered client files to an assistant that answers in 30 seconds during the engagement.
Supporting your clients: from quote to debrief
Four key moments where an in-house assistant transforms the client experience, without inflating the bill.
Accounting firms: three assistants before summer
Doctrine, client, firm — where to start, in what order, at what scale.
Grow your revenue without growing headcount
15 to 25% more sales capacity, without hiring. Four levers, measured across forty mid-sized firms.
Executive AI is not the same as operations AI
Three different readings of the same tool. How to configure for all three without cannibalizing.
Preparing your company for AI assistants
Internal framework, management posture, signals to send. Before the tech, the human framework.
Your SaaS stack, your real hidden costs
When adding up subscriptions underestimates the bill. Method for taking back control.
The hour lost every day
Searching for information, copy-pasting, back-and-forth emails. The silent cost of an under-equipped organization.