Whitepaper
Human–Agent Collaboration: Modes, Rails & the Third Space
Sol–Kim Studio Flagship Whitepaper (v1 – coming 2026)
For many people, AI is becoming a thinking partner.
This whitepaper is our deep, structured answer to a simple question:
What does it actually take to work well with agents – in a way that is useful, safe, explainable, and human?
It distils Sol–Kim Studio experiments into a practical model built around three core ideas:
Modes – the different ways humans work with agents.
Rails – the boundaries, roles and processes that make collaboration safe.
The Third Space – the shared workspace where human intentions and agent patterns meet.
Who it’s for
Leaders and founders who want AI to help their teams, not quietly erode trust, judgement or wellbeing.
Team leads and HR who are being asked to “roll out AI” but can see the mix of enthusiasm, anxiety and uneven skills underneath.
Experienced professionals in day-to-day roles (design, product, comms, strategy, operations and similar) - the people who are actually using tools like ChatGPT, Claude, Copilot and Gemini to get work done.
Consultants & advisors who need clear language and diagrams to explain human–agent collaboration to clients.
You don’t need to be technical. You do need to be curious about how thinking, decision-making and work are changing.
What the whitepaper covers
The paper is organised around a simple arc:
1. The shift: from tools to teammates
We start with what’s actually changing at work:
Why “AI as a teammate” is a deeper shift than “AI as a tool”.
The move from Navigation Mind (“click through tools and docs”) to Agent Mind (“express intents and work in a shared space with an agent”).
The cognitive reality of working with an agent regularly.
2. Four Modes of working with agents
We introduce a vocabulary for how people actually use agents day to day:
Transaction Mode – “Do this task for me.”
Curious Mode – “Help me explore this.”
Builder Mode – “Let’s make something.”
Collaboration Mode – “Let’s run this together.”
Panic as a failure mode, when AI creates more work, not less.
For each mode we outline:
When it’s useful.
Where it breaks.
Simple steps to move into the next mode.
3. Rails: making AI safe, bounded and explainable
We then look at the structures that keep this safe and human:
Boundary Rails – where AI belongs (and doesn’t): ethics, AI-free zones, high-stakes decisions.
Scope Rails – what this agent is responsible for, and what stays firmly human.
Process Rails – how work actually flows: where you chat, where “final” versions live, how decisions are reviewed and logged.
Rails are treated as shared agreements, not just hidden risk controls – something people can see, help design, and actually use.
4. Mindspace, trust and the Third Space
Here we name what many people are already experiencing:
How agents become a place to park ideas, organise thinking, and co-regulate emotion.
How trust with an agent really forms over time (consistency, transparency, feedback loops), and what healthy vs unhealthy trust looks like.
The Third Space as the hybrid workspace where your chats, documents and rituals live alongside the agent’s memory and structure – the place your shared work actually happens.
5. Tiles & pattern libraries – a teachable way to explain “how it works”
We use tiles as a simple way to explain how agents “think”. They don’t store facts like a database, but rather they propose possible next steps based on patterns they’ve seen before - structure, tone, suggested actions.
Collaboration is partly about which tiles you choose to follow. Over time, your Third Space is shaped by the tiles you keep and the ones you retire.
6. Putting this to work: AI literacy as the new digital literacy
The paper finishes with practical starting points for:
Individuals – a simple sequence with Modes vocabulary, a “first 10 messages” ritual, and three basic rails.
Teams – designing one shared Collaboration project with explicit rails and a lightweight weekly “Agent Retro”.
Organisations – moving beyond “everyone has Copilot” to a clear Third Space: tools, rails, rituals and responsibilities.
We treat AI literacy as the new digital literacy: knowing how to talk to an agent, set rails, co-think – and not lose your own voice in the process.
How it connects to the rest of the Studio
The whitepaper sits on top of the Studio’s day-to-day work:
Frameworks outline the core models – Modes, Rails and the Third Space – in simple diagrams and explanations.
Toolkits translate those into things you can use now: quizzes, checklists, canvases and rituals.
Stories has essays, experiments and studio notes from inside our collaborations – the lived reality underneath the diagrams.
The whitepaper ties all of this together into one reference document you can send to a colleague or client who needs a thoughtful, non-hyped overview of human–agent collaboration.
Stay in the loop
Human–Agent Collaboration: Modes, Rails & the Third Space – v1 (coming 2026)
We’ll share draft frameworks, toolkits and launch details on LinkedIn.