Sean Rintel
Sean Rintel Senior Principal Research Manager, Microsoft Research
Brisbane, Australia

Sean Rintel

I study how people collaborate using technology,
and turn those insights into things that matter.

105+ Publications
4,745 Citations
12 US Patents
7× Hackathon Wins
6× Best Paper Awards

I work and live in Brisbane, Australia, on the lands of the Turrbal and Jagera peoples.

Research Vision

The richness of collaboration technology lies in its adaptability to purpose, not its fidelity to reality.

That applies equally to AI: Frontier AI models will improve to equal or exceed some human capabilities, but the next collaboration frontier will be AI that adapts to human intentionality.

For nearly three decades, I have investigated what actually happens when people try to work together through technology. Not the features they use (but the social practices they build around them): how they manage attention, handle trouble, and make meaning together.

This question first drew me to study how strangers open conversations on Internet Relay Chat (1997) and how long-distance couples cope with video-calling glitches (2007–2015). It led me to Microsoft Research, where I have studied hybrid meetings, telepresence robots, avatar communication, and the pandemic's impact on meetings.

In recent years, this line has converged on a new challenge: How should Generative AI support human collaboration, without undermining the thinking that makes collaboration valuable?

This work has moved from communication research through collaboration design to a newer, harder challenge: protecting human cognition from AI's unintended costs. The practical impact is direct, informing Microsoft Teams features, 12 US patents, and the New Future of Work Report series.

More Background

What I work on

I am part of the People-Centric AI team at Microsoft Cambridge UK, developing knowledge, model capabilities, and experiences that enable human agency, support creativity and collaboration, and ensure equitable participation. Previous projects include Tools for Thought project, investigating how to help people think and collaborate better with AI, and Intentional Meetings workstream, investigating how to evolve purposeful meeting systems, behaviours, and cultures with Generative AI.

I run cross-org initiatives, shape strategy, and spend energy making research legible to the people who can act on it, whether that's a product team, an executive audience, or a public one. Our New Future of Work thought leadership has reached well beyond Microsoft, into industry and policy conversations globally.

I am a Partner Investigator on the Australian Research Council Centre of Excellence for Quality Work in the Digital Age (2026-2032).

How I work

My background is in qualitative sociology. Since then I've used and managed a wide range of methods to understand how people's communicative choices interact with designers' technical ones. I'm also exploring how AI can augment research sensemaking, and how vibe-coding is changing research practice. I've mentored researchers, postdocs, and students at all levels, and worked closely with product, design, and policy leads.

Conversation Analysis Ethnography Ethnomethodology Design Research Field Studies Interviews Lab Studies Mixed Methods Surveys Technology Probes Telemetry

Research to Product

I have co-led research that transferred to Microsoft Teams (Companion Mode, Spatial Audio). I've delivered evidence-backed decision presentations and prototypes, and managed initiatives to present those of others, to shape the Teams meetings strategy. I've driven AI-driven experiences both internally and open-source (such as Promptions to help users steer AI). As part of the Tools for Thought project, I've delivered educational presentations on AI usage to multiple organisational groups.

Employment

2014 – Present: Human-Computer Interaction Researcher, Microsoft Research, Cambridge UK.

2010–2014: Lecturer in Communication, The University of Queensland, Brisbane, Australia.

Education

PhD Sociology (Communication), University at Albany, SUNY, 2010. Chaired by Professor Emerita Anita Pomerantz.

MA English, The University of Queensland, 2000.
BA (Hons) English, The University of Queensland, 1995.

Research Map

Technology frames, but does not determine, social action.

1997 – 2015

Communication

The foundation: close analysis of how people actually interact through technology (conversation practices, expressive possibilities, meaning-making). You cannot design good collaboration tools without this understanding first.
2016 – 2021

Collaboration

Building on that understanding, designing systems that support collaboration as a social practice: hybrid meetings, telepresence, spatial audio, avatar presence. The insight: social reality matters as much as objective reality.
2022 – 2026

Tools for Thought

The current challenge: as GenAI accelerates, ensuring AI scaffolds the thinking that makes collaboration valuable (rather than bypassing it). Tools for thought are tools that extend human cognition, not replace it.

View full publications →

Hover over a dot to see details · Click to open the paper

Communication

Before you can design for collaboration, you need to understand how people actually interact.

From IRC to Videocalling

People orient to the expressive possibilities of technology.

My earliest research, studying how strangers manage interaction on Internet Relay Chat (Human Communication Research, 1997), established an approach I have applied ever since: close analysis of how people actually interact through technology, rather than how they say they do.

This conversation-analytic method revealed that even in the simplest text-based chat, people construct sophisticated social practices: openings, turn-taking, and recovery from ambiguous silences. These are not features of the technology but human achievements accomplished through the technology.

The same method revealed how long-distance couples opportunistically use video-calling glitches as relational resources, how domestic soundscapes can be instrumentalised for ambient awareness, and how membership categorisation works as embodied social action.

IRC conversation analysis extract Video-calling conversation analysis extract
Close analysis of interaction, from IRC text (1997) to video-calling couples (2010–2015).
Crisis Memes paper figure Fail Pets paper figure
Crisis memes and fail pets: vernacular genres where people retool templates and brand mascots into expressive, communal resources.
Templates, Mascots, Memes

The same pattern holds beyond one-to-one interaction: people repurpose the expressive affordances of technology to make meaning together.

Crisis Memes (2013) showed that the templatability of image macros is what makes them central to internet culture and freedom of expression: people take a shared visual frame and rework it for their own situation, turning private feeling into public, remixable commentary.

The Evolution of Fail Pets (2012) traced how error screens evolved from terse alerts into whimsical brand mascots (a small design move that turned moments of failure into occasions for empathy, humour, and brand identity).

Across both, the pattern is the same: technology sets the frame, but people decide what the frame means, together.

Design takeaway: study what people do with technology, not just what they say they do. Social practices, not features, are the design unit.

All publications in this area

Collaboration

Collaboration is a social practice, not a feature set.

Remote & Hybrid Collaboration

Before we can design AI that supports collaboration, we need to understand the social practices that constitute it.

People talk collaboration into being. They negotiate who speaks, who's paying attention, and what the interaction means.

Across 30+ papers on meetings, hybrid work, and telepresence, the same patterns recur:

  • Embrace asymmetry in collaboration technologies. In Hybridge (CSCW 2024), giving remote and in-room participants different interfaces (individual displays around the table for in-room, a 3D digital twin with seat-and-viewpoint agency for remotes) closed the co-presence and agency gap that conventional symmetric video calls cannot. Inclusivity comes from tailoring to each endpoint's needs, not from forcing visual sameness.
  • The meeting is not a neutral container. In Meeting (the) Pandemic (CSCW 2022), videoconferencing fatigue was found to arise because tools were designed around steady-state assumptions and a taken-for-granted balance of task and social work. When the rest of the workday collapsed into the meeting, the container broke. Designs must support sociality as well as effectiveness, not assume one is a side-effect of the other.
  • Make space for social time. In Making Space for Social Time (CHIWORK 2022), we showed that traditional videoconferencing skips the conversational transitions (the small talk before, after, and around meetings) that carry collegiality and productivity. The CT-Space prototype used spatial and temporal metaphors, spatial audio, and visible group clustering to restore these transitions without requiring all-day connection. Conversational transitions should be a standard requirement for videoconferencing.
  • Telepresence is as much a social problem as a technical one. A ten-paper program on mobile robotic telepresence at MSR found that workplace adoption fails when robot affordances mismatch workplace social practices. VROOM (CHI 2020) showed that asymmetric mobile robotic telepresence capabilities can enhance the sense of a remote person's belonging to a space.
  • Freelance platforms impose management work on everyone — and account for none of it. Two studies on corporate employees hiring via freelance platforms (Stuck in the Middle (CSCW 2020) and Invisible Work (CSCW 2022)) found that hirers must absorb significant transaction costs the platform was supposed to eliminate, while freelancers perform extensive coordination and relationship work that is structurally invisible to the platform and therefore unrewarded. The management burden on both sides is a feature of the design, not a failure of the people using it.
Hybridge: spatial inclusion for remote participants in hybrid meetings Perspectives: inclusive and equitable hybrid meeting experiences CT-Space: conversational transitions before, during, and after video meetings VROOM: virtual robot overlay for online meetings
Hybridge, Perspectives, CT-Space, and VROOM: prototypes that give remotes spatial presence around the table, restore the conversational transitions that meetings normally skip, and give telepresence robots a meeting-grade face.
(Re)Configuring Hybrid Meetings

Meeting design must be meeting-centred, not user-centred.Best Paper

Hybrid meeting breakdowns arise not from individual UX failures but from mismatches in the meeting ecology: the interplay of practices, technologies, and spaces. This reframing shifts design from fixing individual interfaces to supporting the meeting as a whole.

(Re)Configuring Hybrid Meetings · CSCW 2020

Spatial audio in video meetings

Spatial audio improves turn-taking and social presence.Hon. Mention

In a 75-person experiment, enabling spatial audio led to more equitable turn-taking and higher perceived social presence, because audio spatiality gives listeners the same environmental cues they rely on in person.

Hear We Are · CHIWORK 2023

Multitasking analysis

Multitasking is a feature, not a bug.Hon. Mention

A large-scale analysis of meeting behaviour at a global technology company found that multitasking is pervasive, contextual, and sometimes functional. People switch attention strategically based on meeting relevance, not just distraction.

Large Scale Analysis of Multitasking · CHI 2021

Parallel chat in video meetings

Parallel chat is powerful and perilous.

The text chat alongside video meetings creates a secondary channel that enables inclusion (quiet participants speak up) but also division (side conversations fracture attention).

The Promise and Peril of Parallel Chat · CHI 2021

Avatars & Extended Reality

Realism is not the only axis that matters.

Three studies on avatars and 3D video in meetings show that representation choices shape mood perception, evaluation, and inclusion in ways that pure fidelity does not predict.

  • Cartoon faces can outperform realistic ones over time. In a two-week HoloLens 2 field study (IJHCS 2025), realistic avatars raised expectations their animation could not meet, producing more mood-perception errors than cartoon avatars; words, tone, and movement mattered more than rendering style.
  • Avatar evaluation depends on ecological validity. In AR experiments with simulated facial-animation noise (ANIVAE 2024), higher ecological validity shifted judgement away from noise parameters and toward empathy and gender biases (so "acceptable" avatar quality cannot be claimed without controlling for Context, Culture, and Character, the Triple C).
  • Spatial 3D video wins attention, loses comfort. An Equal Seat at the Table (CHI EA 2024) placed webcam-derived 3D video of participants in a shared virtual meeting room: attention and co-presence improved over the grid, but the grid still felt more comfortable and professional (familiarity is itself a design constraint).
  • Head movement drives avatar effectiveness. In an all-avatar videoconferencing study (Nods of Agreement, CSCW 2025), webcam-driven avatars that nodded naturally produced better meeting outcomes and higher satisfaction than audio-driven or static avatars (natural motion, not rendering fidelity, is the key driver of avatar credibility).
Avatars in mixed-reality meetings: realistic vs cartoon facial likeness Ecological Validity and the Evaluation of Avatar Facial Animation Noise An Equal Seat at the Table: 3D video in shared spatial context
Realistic vs cartoon avatars in mixed reality (IJHCS 2025); avatar facial-animation noise in AR (ANIVAE 2024); 3D video in a shared virtual meeting room (CHI EA 2024).
All publications in this area
Hybrid meetings & remote work
Telepresence robots
Avatars & extended reality
Enterprise freelancing

Tools for Thought

We should ensure that AI scaffolds the thinking that makes collaboration valuable.

The Problem

The features that make GenAI productive can also erode the thinking that makes work valuable.

Generative AI is transforming knowledge work fast. The central question is what AI should do, not what it can do. My research finds a real tension: the features that make GenAI productive (fluency, speed, automation) can erode the critical thinking and intentionality that make human work valuable.

The next generation of AI tools must keep humans in the cognitive loop. That means supporting self-monitoring, not bypassing it.

Design takeaway: AI interfaces must do more than accelerate. Support verification, doubt, and stewardship. Keep the human in the loop.

Tools for Thought Venn diagram: Working With Purpose × Thinking By Doing
The Tools for Thought framing: AI that helps us figure out the job as well as do it.
Critical thinking and GenAI study

GenAI reduces critical thinking effort.87K downloads

In the 2nd most downloaded paper on the ACM Digital Library, we surveyed 319 knowledge workers and found that higher confidence in GenAI is associated with less critical thinking, while higher self-confidence is associated with more. GenAI shifts critical thinking toward verification and stewardship rather than generation. AI interfaces must build in moments to question, not just accelerate past them.

The Impact of Generative AI on Critical Thinking · CHI 2025

Metacognition brain illustration

Metacognition is both the demand and the opportunity.Best Paper

Using GenAI well requires monitoring your own thinking (knowing when to trust output and when to question it). That cost is also an opportunity. Design interfaces that scaffold this reflection: show confidence signals, build in checkpoints, make alternatives visible.

The Metacognitive Demands and Opportunities of Generative AI · CHI 2024

Ironies of generative AI

The "ironies" of GenAI parallel classic automation failures.

Drawing on Bainbridge's ironies of automation, we showed that GenAI can reduce productivity through over-reliance, loss of skill, and misalignment between user intent and AI output. The solution lies in better interaction design that keeps humans in the cognitive loop, rather than in more or less AI.

Ironies of Generative AI · IJHCI 2024

The pervasive problems of ineffective meetings
Lack of intentionality is the persistent meeting problem motivating a research program that follows the meeting lifecycle.
A Proving Ground · Meeting Intentionality

How can Generative AI help people reflect on meeting goals?

Without a solid understanding of why we are meeting, meetings will always be ineffective. Our work on meeting goals is unique in both HCI and Meeting Science, both because of our empirical focus on goals, and because we have now explored all phases of the meeting lifecycle: before, during, and between meetings.

GenAI is well-suited to this. It can surface goals that people haven't articulated (asking questions organisers rarely think to ask themselves). It can prompt reflection at the moment when action is still possible. And it can carry intent across sessions, so goals accumulate rather than evaporate. The risk is that the same fluency that surfaces intent can also substitute for it. Our designs treat GenAI as a prompt for human thinking, not a replacement.

There's also an AI flywheel here. Explicitly expressed goals have direct value: clearer plans, sharper meetings, better follow-through. They also feed AI systems with structured intent that can be read and acted on across sessions (rather than reconstructing purpose from scratch each time).

Mental models of meeting goals

Foundations · Mental Models of Meeting Goals

The formative study underpinning the program: how organisers and participants mentally model meeting goals, and what intentionality support would have to look like to fit those models.

Mental Models of Meeting Goals: Supporting Intentionality in Meeting Technology · CHI 2024

Field experiment overview: 361 employees, 7196 meetings

Before meetings · Nudging Attention to Workplace Meeting GoalsCHI 2026

A preregistered field experiment with 361 employees across 7,196 real meetings. Even a brief pre-meeting goal-reflection prompt improved self-reported awareness and behaviour.

Nudging Attention to Workplace Meeting Goals · CHI 2026

What Does Success Look Like? Prospective reflection prototype

Before meetings · What Does Success Look Like?

An AI-assisted prospective reflection tool that helps organisers articulate success conditions before a meeting begins, catalysing intentionality at the moment plans are still malleable.

Catalyzing Meeting Intentionality with AI-Assisted Prospective Reflection · CHIWORK 2025

CoExplorer adaptive meeting interface

Before & during meetings · CoExplorer

A GenAI-powered adaptive interface technology probe spanning planning and running video meetings. Used as a generative design vehicle to surface what intentionality support could look like across the lifecycle.

The CoExplorer Technology Probe · DIS 2024

Are We On Track? AI-assisted goal reflection

During meetings · Are We On Track?

Active and passive AI-assisted goal reflection helps participants recalibrate mid-meeting without disrupting flow. Deployed as a technology probe inside Microsoft.

Are We On Track? AI-Assisted Active and Passive Goal Reflection During Meetings · CHI 2025

Chain of intentionality across meetings

Between meetings · Temporal Work Across Meetings

Interfaces that use Generative AI to scaffold the connective work people do between meetings: tying threads forward, surfacing unresolved goals, and making the gap between sessions productive rather than lost.

Designing Interfaces that Support Temporal Work Across Meetings with Generative AI · DIS 2025

All publications in this area
Meeting intentionality
AI, critical thinking & cognition
New Future of Work reports
Innovation & Translation

Research that reaches products, patents, and practice.

12 US patents · 7 Microsoft Hackathon wins (1 Grand Prize, 5 Firsts, 1 Second).

Product Feature

Microsoft Teams Companion Experiences: Meetings are easier when your computer and phone can work together

Teams meetings used to force a choice: join from your computer or your phone, not both. Companion Experiences removed that constraint. Join on your PC and use your phone to share live video, show your screen, remotely advance slides, or move your audio and camera between devices (without conflicting audio streams or a second meeting join). How to join on a second device →

What it does and how it got there
Presenting live video from a phone in a Teams meeting Sharing the phone screen in a Teams meeting Remotely controlling PowerPoint from a phone

Present live video or photos. Remote participants get a view of your physical surroundings (a whiteboard, a product, a space) by joining your phone as a companion device and sharing live mobile video or photos displayed on your computer screen.

Present your phone screen. Share mobile apps or phone-only content directly into the meeting, or project your phone locally to the meeting room without remote participants seeing it.

Remote-control PowerPoint. Advance slides from your phone while presenting from your computer (useful when you need to move around the room). Multiple presenters can independently review the deck and request control, cutting the time gaps in handoffs during Q&A.

Switch devices without rejoining. Toggle your video, speaker, and microphone between devices in any combination as you move. No disconnect, no second audio stream.

The project began as Skype Unleashed, a 2015 Microsoft Hackathon entry (a vision video showing what calling could look like if a phone and PC worked in concert rather than competing). It won the Business category. The team spent two years building a working prototype called Project Wellington, which joined Brian MacDonald's early Microsoft Teams incubation. Features were announced at Build 2017 and Enterprise Connect 2018, where Teams took the top prize.

The Companion Experiences team
"I was showing the whiteboard through the phone to the remote participants. It genuinely helped the course of the meeting." — L., Microsoft UK Marketing and Operations

The multi-device meeting research behind this project contributed to three US patents: Companion Devices for Teleconference (granted 2022), Enhanced Joining Techniques for co-located device detection (granted 2021), and Proximate Resource Pooling for AV resource sharing across nearby devices (granted 2017).

Open Source Feature

Promptions: If You’re Building UI for AI, Give Users the Power of Choice

Most AI interfaces put the burden of exploration on the user: write a prompt, get a response, guess at a better prompt. Promptions inverts this (it generates contextual UI elements: radio buttons, checkboxes, toggles) so users can explore variations of the AI's response without rewriting anything.

It is open source: clone the repo on GitHub or see more on AI Foundry Labs.

How it works and what it’s built on
Promptions compared to standard AI interface — side by side

The standard AI interaction loop puts all the steering work on the user: you write a prompt, interpret the output, then try to refine your wording to get closer to what you wanted. A recurring frustration in user research: "I have to do so much work around my prompt to give it the context, point it to the tone, and a lot of times I'm repeating those same things."

Promptions breaks that loop. When you submit a prompt, the system generates a small set of contextual options alongside the response (controls specific to what you just asked, not generic settings). Adjust a toggle or pick a radio button and the response updates immediately. Users in early studies found it surfaced possibilities they hadn't considered: "It generated options that were things I would maybe assume the system couldn't handle."

Promptions system model: option module, rendering engine, chat module

Under the hood: an Option Module ingests the prompt and conversation history and outputs a structured set of options. A rendering engine turns those into inline UI controls. The Chat Module incorporates the selected options as grounding when generating or regenerating the response. Changing a control reruns the Chat Module immediately (no re-prompting required).

Promptions is grounded in CHIWORK 2025 research on dynamic prompt middleware. It ships as a TypeScript monorepo for React and Fluent UI applications with OpenAI integration. Invented by Ian Drosos (past Microsoft Research Resident) with Jack Williams, Advait Sarkar, Nicholas Wilson, Payod Panda, and Sean Rintel (a collaboration between the Tools for Thought and ENCODE projects at Microsoft Research Cambridge).

12 US Patents

These patents and prototypes trace directly from research insights: meeting-centred design → spatial hybrid meeting systems; avatar and telepresence research → personalised animation and companion devices; robotic telepresence → multi-endpoint mixed reality.

Hybrid Meeting Systems

  • Collaborative System: spatial hybrid meetings with avatar-equipped displays and digital twinsgranted 2025
  • Established Perspective UI: spatially consistent seating in video meetingsgranted 2023, 2025
  • Simulated Choral Audio Chatter: crowd audio for virtual event atmosphere2024

Extended Reality & Avatars

  • Multi-endpoint Mixed-Reality Meetingsgranted 2022, 2024
  • Headset Virtual Presence: headset-controlled avatars with emotional expressiongranted 2023
  • Personalized Avatar Animation: few-shot ML model for audio-driven avatar movement2025

Interaction & Input

  • Computing Device Headset Input: headphones as a general-purpose input devicegranted 2023
  • Semantic User Input: context-aware gesture-to-action mapping2023
  • Mixed Reality Workflow Generationgranted 2025

Multi-device Collaboration

  • Companion Devices for Teleconferencegranted 2022
  • Enhanced Joining Techniques: co-located device detectiongranted 2021
  • Proximate Resource Pooling: AV resource sharing across nearby devicesgranted 2017

7× Microsoft Hackathon Winner

Promptions hackathon
2024
Promptly 2nd place

Dynamically-generated UI for AI: everyday, any purpose, anywhere. Now Promptions on AI Foundry Labs.

Inclusive Avatars in Mixed Reality
2021
Inclusive Avatars in MR Grand prize

Teach your avatar to move like you, using few-shot training.

LinkedIn Career Classroom in Microsoft Teams
2019
LinkedIn Career Classroom 1st place

Connecting children from anywhere to career role models.

MILC — Mobile Intelligent Lenses
2019
MILC 1st place

Mobile Intelligent Lenses for everyday Mixed Reality at work.

TACTILES hackathon — tangible cloud intelligence
2017
TACTILES 1st place

Tangible Active Cloud Intelligence: making the cloud a conversation.

Skype Unleashed / Companion Experiences
2015
Skype Unleashed 1st place

Surface Hub cross-device patch-panel UI. Productized as Companion Mode in Microsoft Teams. Garage Wall of Fame.

What's Next

Frontier AI models will change, but the next collaboration frontier will be AI that supports human intentionality.

Generative AI is now in every collaboration tool. Most of it is built around a single idea: acceleration. My research suggests a different question.

Instead of asking "what did the AI do for you?", we should ask "did the AI help you think more clearly about why you're doing this?"

Design Principles

  1. Treat the medium as a participant, not a pipe. Communication technologies shape what counts as presence and contribution. Study the social practices that emerge from them, rather than assuming the channel is neutral.
    • The AI turn is a real conversational turn. When AI surfaces a suggestion, flags a drift, or generates a summary during a meeting, it takes a turn. Unlike a silent notification, AI speech creates social pressure and disrupts floor dynamics. Treat AI utterances as conversational contributions that need turn-management, not just information displays.
    • AI narration is a social act, not neutral documentation. When AI produces a summary, transcript, or action-item list, it creates an authoritative account that shapes accountability. Some contributions get foregrounded, others omitted. This is a narrative act. Design for visible narration choices and let participants co-author the AI's account of what happened.
    • AI uptake reproduces who counts as a speaker. ASR and language models trained on majority speech systematically under-transcribe accented, overlapping, or quiet voices. When AI selectively hears certain people, the medium shapes who counts as a participant. Audit AI uptake with the same rigor you apply to interface accessibility.
  2. Embrace asymmetry. People come to meetings differently (different bodies, devices, situations, power). Equal participation requires different affordances, not identical ones.
    • AI models the meeting from a single vantage point. Most meeting AI produces one account as if no one in particular made it. But the meeting is experienced differently by the host, remote participants, and those with different power. A single AI account privileges one perspective while appearing objective. Design for AI that can be perspectival or surface multiple accounts, not just one synthetic view.
    • Personalization ≠ individualization. AI that learns your preferences adapts to past behavior, not structural position. A remote participant's needs differ from a collocated one not by taste but by role and access. Design AI adaptation at the level of role and context, not just behavior patterns.
    • Differential AI access amplifies structural inequity. When some have AI assistance and others don't, the meeting's information gap widens. AI amplifies existing privilege rather than leveling it. Design for AI as a meeting resource, not a per-user premium.
  3. Design for intentionality and process, not just output. Help people think about why before optimising how. Speed without purpose produces faster noise. Goals should persist across sessions as named artifacts (legible to people and AI alike).
    • Surface the intent layer before the generation layer. AI that jumps to output skips the most valuable step: making intent explicit. Require users to articulate why they want something before AI generates. This is cognitive work worth supporting, not friction to eliminate. Users who author their intent get more useful outputs and own them more.
    • Distinguish AI-assisted articulation from AI-substituted articulation. AI that helps you find words for what you intend is different from AI that proposes new goals. The first is assistance. The second trains people to accept AI-generated intent. Design for assistance as the primary mode. Treat AI-generated intent as scaffolding to own and revise, not silently accept.
    • Treat goals as persistent artifacts, not session-scoped state. AI that forgets meeting goals when the session ends treats intent as ephemeral. Goals are durable commitments. Keep them across sessions, visible to all, so AI can refer back to them over time.
  4. Scaffold metacognition and calibrate confidence. Using GenAI well requires monitoring your own thinking. Knowing when to trust output and when to question it is genuinely hard (and a real design opportunity). Build interfaces that help users calibrate trust and resist fluency-as-authority.
    • Fluency is a Trojan horse for authority. GenAI produces polished output regardless of accuracy. People equate well-written with correct. Design must counteract this. Surface where claims come from. Mark uncertainty structurally, not just with disclaimers. Don't present AI output as authoritative.
    • Design for productive scepticism, not just informed consent. Disclaimers and confidence scores only inform consent. They don't foster active doubt. Create moments that make people think: ask them to predict before seeing AI output. Flag where it changed from last time. Add verification prompts for high-stakes outputs.
    • The verification burden is a design failure, not a user responsibility. When verification falls entirely on users, it disadvantages those with less expertise or time. Make verification tractable: link claims to sources. Distinguish high-stakes outputs. Don't ask people to verify the unverifiable.
  5. Enable thinking by doing. Insight rarely arrives before action; it emerges through making, trying, and revising. AI tools should treat externalisation as the point, not the preamble: lower the cost of making rough, provisional things so people can think through artifacts rather than only about them. Reversibility matters more than first-try fluency.
    • Treat the first AI output as a provocation, not a draft. "Generate, then edit" treats AI output as 80%-done. But seeing a response anchors thinking to its structure. Start with alternatives, questions, counterproposals. Open possibility space before closing it. AI's first job is discovery, not production.
    • Make AI outputs structurally reversible. If AI enables cheap experimentation, make undo and branching cheap too. Design AI interactions as explorations with clear return paths. Make committing harder than exploring.
    • Design AI for dialogue about the task, not just production of outputs. The point isn't "AI writes it." It's "AI helps me think about what it should argue or exclude." Support dialogue about the task before generation. Ask what we're trying to do, what success looks like, what we're not doing.
  6. Design for meaningful shared attention. A firehose of notifications is not collaboration support. As AI becomes a genuine participant in shared work, interfaces must cultivate the conditions for sustained attention rather than competing for it.
    • Distinguish AI-generated urgency from human-generated urgency. AI notifications create synthetic urgency that can override negotiated priorities. A human raising a point exercises judgment. AI flagging drift executes code. Design for AI that reads the room before intervening, not just its own relevance.
    • The attention unit is the meeting, not the individual. AI optimized for individual attention fragments shared cognitive space. Design at the meeting level: what does this group need to see, not what does this person need.
    • Design for attentional recovery, not just attentional capture. Most AI features compete for attention. The harder problem is helping people return to shared focus after a drift or tangent. AI that senses divergence and scaffolds re-convergence beats AI that just alerts.