Communication systems for human–AI collaboration.
The bottleneck isn't AI. It's communication infrastructure.
Organisations have deployed AI tools. Teams use them differently. Output quality is inconsistent. Communication protocols haven't caught up.
The bottleneck isn't technology — it's the absence of communication infrastructure between people, between people and machines, and between machines.
And the problem starts earlier than most organisations realise. People's assumptions about AI — what it can do, what it should do, what it will do wrong — shape every interaction before a single word is typed. Without structure, input drifts toward the generic. Output follows.
I design communication systems that make human–AI collaboration predictable, accountable, and scalable. Three channels, one integrated approach:
H2H Human ↔ Human — how teams communicate about and around AI-assisted work
H2AI Human ↔ AI — how people brief, steer, delegate to, and review AI systems
S2S System ↔ System — where AI systems coordinate, hand off, and need human oversight
The Triangle (Situation / Purpose / Form) — governs every exchange. Front-loads the shared understanding that prevents rework.
The Four Layers (Context → Dialogue → Delegation → Review) — governs the collaboration lifecycle. Clarify, iterate, delegate, verify.
Grounded in distributed cognition, joint intentionality, grounding theory, and sociotechnical design. Two decades of applied communication expertise.
One methodology. Three formats. Each matches the buyer's level of authority and commitment.
Same method. Different surface. Different depth. Some people start by experiencing it, some by practising it, and organisations by installing it.
Not every situation fits a format. Some organisations are mid-transition and need someone who can see the full picture — structural, cognitive, human. Some leaders need a thinking partner who understands how communication breaks down when AI enters the picture, and can work with them directly to build what the situation requires.
This is where the full depth of the method meets your specific problem. No fixed scope. No predetermined deliverables. No programme to complete. Just sustained, tailored work on the communication infrastructure your organisation actually needs.
I work in three languages across organisational, leadership, and systems contexts. I've spent two decades in cross-cultural communication and the last several years studying — and designing for — how human work changes when AI participates. If what you're dealing with is structural and communicative, I've probably thought about it.
Tell me what's happening with communication in your organisation. I'll tell you whether this fits.
Start a conversation →+31 6 1764 4141
The Netherlands · Dutch, English, German
90-minute live session
A 90-minute working session where you experience what changes when communication with AI is structured rather than improvised.
The session begins by surfacing your operative assumptions: what you expect AI to do, what you're concerned about, and what's actually at stake in your work. This calibration ensures that the practice rounds start from examined assumptions rather than default ones.
You then run the same task three times, adding one structural element each round: context (Situation), intent (Purpose), and constraints (Form). The improvement is visible and immediate. The final round demonstrates that structured communication front-loads shared understanding — eliminating the rework loops that consume hours of professional time.
The natural tendency in AI interaction is to converge toward the machine's register — producing cleaner, more generic, more predictable input. This isn't a failure of sophistication. It's an interactional gravity that operates in the absence of structure.
The session makes this visible and provides the counterweight: a method (Situation / Purpose / Form) that maintains signal fidelity under conditions that naturally degrade it.
By the end of the session, you will have experienced three things: that AI output quality is a function of input structure, not AI capability; that your own assumptions about AI shaped your input before you typed a word; and that a simple communication framework produces consistently better results than intuition alone.
Meet Your AI builds awareness. If you want the complete method — the full collaboration lifecycle from context-setting through delegation and review — the next step is Onboard Your AI, a half-day workshop that develops the complete working method.
Onboard Your AI →Half-day workshop — the complete method
A half-day workshop that builds the complete human–AI collaboration method. Four modules covering the full cycle: establishing shared understanding, building through dialogue, delegating with boundaries, and reviewing responsibly.
Where Meet Your AI demonstrates the shift, Onboard Your AI builds the working infrastructure. You leave not with awareness but with a repeatable personal method for every AI interaction — one that counteracts the natural tendency to flatten input toward what the machine rewards.
How to construct the context that AI systems cannot infer. You learn to front-load shared understanding rather than relying on iterative repair. The module makes visible what happens when context is missing: AI fills the gap with statistical averages, producing output that sounds competent but misses the point.
How to build collaboratively through structured exchange rather than one-shot prompting. You practice iterative refinement where each turn adds signal rather than repeating instructions. The module addresses the common pattern of abandoning output too early or accepting it too readily.
How to assign outcome responsibility with explicit boundaries and checkpoints. You define what AI executes, what AI proposes, and what humans decide. The module addresses the ownership ambiguity that produces both over-reliance and under-use — two sides of the same unclear delegation.
How to verify output efficiently through pattern-based assessment matched to risk. You develop review methods calibrated to consequence: light verification for low-stakes output, structured review for high-stakes decisions. The module builds the habit of examining not only whether the output meets specifications but whether the specifications themselves were adequate.
Partner critique runs through every module. You learn from each other's blind spots — including the assumptions about AI that shape input quality before any method is applied. Someone else's review of your AI interaction reveals gaps that your own eyes systematically miss, because AI's linguistic fluency satisfies your sense that communication has succeeded when it hasn't.
The building blocks philosophy means you assemble what you need rather than following templates. The method adapts to different work contexts, different risk levels, and different collaboration patterns. This is personal infrastructure, not a script.
Onboard Your AI builds individual and small-team capability. It stops explicitly at the individual boundary. Organisational change — protocols, role matrices, accountability structures — requires the Communication Systems Intensive.
Participants who see the method working in their own practice start asking a different question: "How do we make this standard across the team?" That question is the entry point for the Communication Systems Intensive.
Communication Systems Intensive →1–2 day working session for leadership & core teams
This is not training. It's a hands-on reset of how communication actually works across people, AI, and systems.
In 1–2 days, we calibrate, diagnose, design, and test communication protocols inside your real work. Your team leaves with decisions, not homework.
We map and redesign communication across three channels. Most organisations only think about the first.
H2H Human ↔ Human — How teams brief each other, share context, make decisions, and coordinate around AI-assisted work.
H2AI Human ↔ AI — How people frame requests, steer output, delegate tasks, and verify results with AI systems.
S2S System ↔ System — Where AI systems hand off to each other, where automated decisions need human oversight, and where system coordination creates blind spots.
Before any diagnostic begins
Human–AI interaction is shaped by conditions that precede the interaction itself: what people expect AI to do, what they're concerned about, what they assume about how it works. These operative assumptions systematically affect how protocols are activated.
Phase 0 surfaces these activation conditions across the team. Participants identify: the constraints in play (stakes), the expected outcomes, and the risk conditions. This is diagnostic, not reflective — comparable to a pre-flight checklist that confirms conditions are known before procedures are activated.
The exercise produces a shared inventory of operative assumptions that is referenced throughout subsequent phases. It is repeated at the session's close.
We map how communication really happens — not how people think it happens. Through rapid stakeholder interviews, live workflow walk-throughs, and review of real outputs, we build a clear picture of where meaning breaks down.
We translate morning findings into explicit communication rules. Working in real time with the team, we design operational protocols that fit their tools, culture, and risk tolerance.
Optional but recommended
Teams apply yesterday's protocols to current work. Real projects. Real AI use cases. Real decisions on the table. We stress-test for clarity, speed, and failure points. Protocols are refined on the spot — not in a follow-up document nobody reads.
The final session is for decisions. What becomes mandatory. What stays optional. How these protocols roll out beyond the room.
Fewer clarification loops. Because briefing standards mean people say what they mean the first time — to each other and to AI.
More reliable AI output. Because structured context-setting produces consistent results instead of random quality.
Clear human vs. machine ownership. Because delegation rules make explicit what AI decides and what humans own.
Faster decisions. Because escalation paths mean uncertainty gets resolved, not circulated.
Less managerial firefighting. Because protocols handle the routine, freeing leadership for what actually needs judgment.
And crucially: a shared language for how work gets done.
For organisations that want ongoing support after the Intensive. Quarterly protocol reviews, adoption monitoring, capability updates as AI tools evolve, and two advisory sessions per month.
Not a trailing engagement. A standing relationship for organisations where communication infrastructure needs to evolve with the technology.
Tell me what's happening with communication in your organisation. I'll tell you whether this fits.
Start a conversation →+31 6 1764 4141
The Netherlands · Dutch, English, German
The theoretical foundations behind the work.
The methods described on this site are not invented from practice alone. They are grounded in a body of research spanning distributed cognition, communication theory, sociotechnical systems design, and the emerging science of human–AI interaction.
This page exists for those who want to see the intellectual infrastructure behind the practical work — whether you are evaluating the approach, considering collaboration, or working in a related field.
Conference paper · CSCW 2026 (in preparation)
This paper argues that when AI systems participate in organisational knowledge work, communication must be treated as infrastructure rather than skill. It proposes a design framework comprising a three-element interaction model (Situation, Purpose, Form) and a four-layer collaboration lifecycle (Context, Dialogue, Delegation, Review), and describes its operational instantiation in a facilitated organisational intervention.
The framework draws on distributed cognition (Hutchins), joint intentionality (Tomasello), situated action (Suchman), actor-network theory (Latour), grounding theory (Clark), and research on pre-communicative conditions in human–technology interaction (Goffman, Turkle, Nass & Reeves). It identifies pre-communicative staging and asymmetric convergence as structural factors shaping interaction quality prior to and during protocol activation.
[PDF will be available here upon submission.]
The work integrates seven research traditions, each contributing a distinct analytical lens:
Cognition as a property of systems, not individuals. When AI participates in work, the cognitive system extends across human and machine — requiring explicit coordination infrastructure.
Collaborative action requires shared goals and mutual awareness of those goals. AI systems cannot form intentions, but humans can structure interactions to achieve functional alignment.
Shared understanding is an active accomplishment, not a precondition. The Triangle functions as a grounding mechanism — front-loading evidence of communicative intent to reduce costly repair cycles.
Plans are resources for action, not determinants of it. Protocols must be designed for adaptation under real conditions, not compliance under ideal ones.
AI systems are actants that shape workflows whether or not they are formally recognised. Communication infrastructure must account for their participation explicitly.
Single-loop learning optimises within existing assumptions. Double-loop learning examines the assumptions themselves. The Review layer builds double-loop capacity into human–AI collaboration.
Interaction is shaped by conditions that precede it: frame-setting, impression management, projection, and automatic social responses to computational systems. These conditions must be surfaced, not assumed away.
Pascal du Ry's work sits at the intersection of communication theory, cognitive science, and organisational systems design. His research focuses on the structural conditions required for effective human–AI collaboration — treating communication not as a competence to be trained but as infrastructure to be designed, installed, and governed.
Two decades of applied communication expertise across multilingual and cross-cultural contexts (Dutch, English, German) informs both the theoretical framework and its practical implementation.
For academic collaboration, speaking, or research enquiries.
Send a message →+31 6 1764 4141
The Netherlands