- Members are polite, guarded, and focused on acceptance
- Purpose, roles, and expectations are unclear
- High dependency on the leader for direction
- Excitement mixed with anxiety about what lies ahead
- Little productive output - energy spent on orientation
- Provide clear direction, structure, and purpose
- Establish team norms, ground rules, and ways of working
- Facilitate introductions and relationship-building
- Be highly available and visible
- Set near-term, achievable early wins to build confidence
- Team members unfamiliar with AI capabilities and limitations
- AI role and boundaries not yet defined
- Caution and curiosity co-exist - people testing what AI will do
- Define clearly what the AI is for - and what it is not
- Establish approved tools and basic governance upfront
- Run low-stakes demos to reduce fear and build familiarity
- Interpersonal conflict and power struggles emerge
- Dissatisfaction with leadership or direction
- Subgroups and cliques begin to form
- Roles and responsibilities are contested
- Performance dips - the most uncomfortable phase
- Name and address conflict early - don't suppress it
- Facilitate resolution through structured dialogue
- Clarify roles, decision rights, and accountabilities
- Maintain direction without becoming autocratic
- Recognise that Storming is a healthy and necessary stage
- Resistance to AI involvement - fear of replacement or deskilling
- Disagreement about how much to trust or rely on AI output
- AI can itself act as a Storming trigger by surfacing inconsistencies
- Address resistance openly; don't dismiss it as irrational
- Clarify AI decision boundaries and human override protocols
- Use AI to reduce conflict about data accuracy, not to impose conclusions
- Conflict reduces; shared norms and values take hold
- Trust builds and collaboration improves
- Roles become clearer and more accepted
- Team develops its own identity and ways of working
- Performance begins to climb toward its potential
- Step back from directive leadership - facilitate rather than control
- Reinforce positive behaviours and recognise progress
- Encourage peer accountability alongside leader accountability
- Introduce more complexity and stretch challenges
- Formalise the norms that are working
- Team begins to integrate AI into normal workflow patterns
- Informal norms emerge around how AI is used day-to-day
- Comfort increases but governance may still be informal
- Formalise AI usage norms - Acceptable Use Policy if not yet in place
- Embed AI into standard processes rather than keeping it ad hoc
- Review early use cases: keep what works, retire what doesn't
- High levels of autonomy, trust, and interdependence
- Problems are solved collectively without leader direction
- Strong shared identity, purpose, and motivation
- Roles are flexible; people step up where needed
- Output is consistent, high-quality, and self-sustaining
- Delegate fully and trust the team's judgement
- Focus on removing blockers and protecting the team's environment
- Maintain stretch - complacency is the main risk at this stage
- Recognise and celebrate sustained high performance
- Begin succession thinking to sustain performance over time
- AI is fully embedded - treated as a reliable team contributor
- Human–AI handoffs are smooth and well-understood
- Team proactively seeks new AI use cases to extend capability
- Maintain oversight - high AI confidence can mask drift or error
- Introduce regular AI output audits to preserve quality standards
- Identify the next frontier - what AI challenge would stretch this team further?
- Team disbands - project ends, members move on, or structure changes
- Emotional responses range from pride to grief and anxiety
- Knowledge and relationships risk being lost if not captured
- Team identity begins to dissolve
- Attention shifts to what comes next, not current tasks
- Acknowledge the ending - don't minimise the emotional dimension
- Capture lessons learned, decisions, and institutional knowledge
- Celebrate the team's contribution and collective achievement
- Where appropriate, deliberately reset to Forming for the next cycle
- Actively hold performance at peak if the team is continuing with new scope
- AI configurations, prompts, and workflows risk being lost at dissolution
- Uncertainty about whether AI tools carry over to the next team
- Document AI configurations, effective prompts, and governance decisions
- Transfer AI asset ownership formally - don't let institutional knowledge disappear
Teams do not move smoothly through phases in sequence. Disruptions - new members, leadership changes, scope shifts, or introducing AI - can return a team to an earlier phase. This is not failure; it is how development works.
Performance genuinely drops during Storming. Teams that skip conflict resolution or suppress disagreement tend to get stuck in a low-level Storming state indefinitely. Working through it is the only path forward.
Despite regression and dips, the overall arc should be upward. Each cycle through the phases - when well-led - typically reaches a higher performance ceiling than the last. The aim is not to avoid regression but to recover from it faster.
Adjourning is not the end. Teams that survive transition move into a new Forming phase - often at a higher baseline. Leadership's role in Adjourning is either to hold the performance gain or to deliberately reset the team with a clear mandate for the next cycle.
New team members and new technology partnerships both start here. Whether it is onboarding a new IT hire, integrating an acquired firm, or introducing a new managed service provider - set clear expectations, establish norms early, and provide visible leadership from day one.
Common in technology change programmes, supplier transitions, and AI adoption. When fee-earners challenge new tools or the IT team debates delivery priorities - name the conflict, hold the structure, and work through it. Suppressing it delays progress. AI tools frequently introduce Storming by surfacing inconsistencies in existing processes.
The IT team and its stakeholders develop shared understanding of how technology change is planned, governed, and communicated. An AI Acceptable Use Policy, clear change management processes, and embedded service desk routines are all markers of a Norming function.
The IT function operates as a proactive business partner. Delivery is consistent, stakeholders trust the team, and innovation happens within a governed framework. At this stage, the conversation with the SLT shifts from "what can IT do?" to "what should we do next together?"
Project completions, leadership changes, or firm restructuring all trigger Adjourning. For the IT function, this also includes supplier exits, platform retirements, and team reorganisations. Capture what worked, celebrate the contribution, and reset deliberately - not by accident.
In traditional teams, humans hold final authority. In high-functioning human–AI teams, the AI's output may be of higher confidence in certain domains than the human's intuition. Teams need explicit protocols for when to defer, override, or challenge AI recommendations.
AI participants can hold multiple functional roles simultaneously - analyst, drafter, reviewer, and summariser at once. This challenges traditional role clarity. Teams must define what the AI is doing at each stage, or risk confusion about who is accountable for output quality.
Human and AI contributions become intertwined in ways that are hard to separate. Authorship and decision ownership become ambiguous. This is not inherently a problem, but governance frameworks must address it - particularly in a legal environment where auditability and professional accountability matter.
High-performing human–AI teams often develop unofficial interaction norms - prompt patterns, verification habits, trust thresholds - before formal policies catch up. Leaders should surface and formalise these early rather than waiting for a governance review cycle.