Your teams already use AI — quietly, without method, without a safety net. We redesign how work gets done so AI doesn’t just assist your team: it joins it, like a real colleague. Visible, reliable, working for the team. And you keep it.
Social AI Lab · Paris · 20 years shaping the future of work (Emakina, BlueKiwi, Anaplan) · For impact-driven organizations
You're not behind. You bought the licences, ran the workshops, maybe even built a few automations. And yet AI still feels like a side project, not an advantage. Because most teams fall into one of three traps.
"Everyone should be using AI."
Leadership announces an AI initiative. A town hall, a Slack channel, then… nothing changes. Adoption stays scattered and fizzles within weeks.
Why it fails: no structural change. You're asking people to change their habits without changing their environment. The corporate equivalent of a gym membership bought in January.
"Let’s add AI to our existing process."
You insert AI into one or two steps of a human-designed workflow. It drafts an email here, summarises a meeting there. Output is generic, sometimes wrong. Trust erodes.
Why it fails: the workflow was built around human strengths — context, intuition, tolerance for ambiguity. AI has none of those. A new engine in a car built for a different drivetrain.
"We’ve trained everyone on the tools."
Individuals get faster. Personal productivity goes up. But the organisation doesn't change. No new capability emerges, no role is redesigned.
Why it fails: training ten people on tools gives you ten slightly faster people. That's multiplication. Welcoming AI as a full participant is addition — capacity that didn't exist before. A very different equation.
If any of these sound familiar, the problem isn't AI. It's the system it's operating in.
Using AI is multiplication: faster people, but the same threads held in the same head. Welcoming an AI colleague is addition: entire stretches of work leave your mental load and become someone else’s responsibility.
The market talks about "hiring" AI. We talk about welcoming it. That’s not just softer language: the way you bring AI in decides whether your teams will trust it or quietly sabotage it. You don’t impose a colleague. You make room for one.
Most organisations are stuck at stage 1 or 2. Here's the full path.
| Stage | Today | Tomorrow |
|---|---|---|
| 1. Shadow | AI runs in secret | The AI colleague is welcomed, visible |
| 2. Tooled-up | People can write prompts | People can delegate and verify |
| 3. Paired | Everyone tinkers alone | Everyone pilots their AI colleague |
| 4. Team | Isolated pairs | Pairs that coordinate |
| 5. Organisation | Scattered uses | An organisation that orchestrates its pairs |
This isn't a metaphor. It's a method. Define the role. Design the interfaces. Onboard the colleague. Measure the output. Iterate.
Every workflow has this shape. The question isn’t "should we use AI?" — it’s "which stages are we still doing by hand that we don’t need to?"
| Stage | AI potential | Human role | Key question |
|---|---|---|---|
| Intake & triage | High | Set the rules, handle exceptions | Can the routing logic be articulated? |
| Research & context | High | Judge relevance, flag gaps | Is the data structured and accessible? |
| First draft | Medium | Set the quality bar, give examples | Can "good enough" be defined by example? |
| Review & iteration | Medium | Apply taste, catch edge cases | Where does judgement override pattern? |
| Decision & trade-offs | Low | Own the decision, manage the relationship | Does this need trust, politics, novel judgement? |
| Delivery & follow-up | High | Handle escalations | Is the delivery format standardised? |
We don't install surveillance. No telemetry, no reading your teams' conversations. We start from what people choose to share — their outputs, their workflows, their rituals. Analysis always happens at team level; individual data belongs to the individual alone. A colleague you welcome, not an eye you impose.
AI Product Studio
From strategic framing to deployment, we design AI-powered products and services — built for your beneficiaries, not for the tech. For organisations that want to create, not just adopt.
AI Adoption Coaching
The tool doesn't make the use. We help your teams understand, experiment, and welcome their AI colleague into daily work. No endless slides — hands-on practice, measured results, confidence. Entry format: a 5-week bootcamp to build your pair.
Social AI Lab
We run experiments on emerging AI uses for inclusion and autonomy — including everyday assistive robotics. A space to imagine tomorrow's AI, built for those who need it most.
Whichever door you take, the goal is the same: your team owns the system, not us. You keep your AI colleagues, your playbooks, your data. You switch tools whenever you want.
Jikken runs on a small team of humans and agents. When you talk to us, you see the model we're proposing, live. And our conviction isn't a slide: after 20 years shaping the future of work, we're putting AI to work for a different ambition — that no one gets left behind by technology. That's the purpose behind our impact products and our Social AI Lab.
See how we run Jikken with our AI colleagues →Christophe Routhieau, founder. Co-founder of Emakina and BlueKiwi Software, contributor to Anaplan's growth. I've led product portfolios, driven transformations, and learned one thing: human adoption is what separates a tool from a lever for change. Today I help impact-driven organisations — NGOs, nonprofits, purpose-led businesses — design and welcome AI that's useful, ethical, and concrete. Jikken — "experimentation" in Japanese — was born from this belief: technology alone changes nothing, humans alone aren't enough anymore. It's the right combination of both that creates real impact.
Named testimonials + LinkedIn links, to collect from the first cohort.
A 30-minute call to see where you stand and where to start. No slides. Just the useful conversation.
Book a call