Workshops: Agentic Engineering Practices

Pair & Mob Programming with an AI Agent: Team Workshop

Most developers use coding agents alone, so good prompts, tricks and warning signs stay in individual heads. In this half-day workshop your team tries pair and mob programming with an AI agent as an extra team member: clear roles, short rotations and a shared screen, so everyone learns from everyone while the agent does the typing.

Who is this workshop for?

  • Development teams starting with coding agents who want to learn together
  • Teams that already practise pairing or mob programming
  • Agile coaches and Scrum Masters looking for new team formats
  • Tech leads who want agent skills spread evenly across the team
  • Mixed teams of newcomers and experienced developers

What you will take away

  • Run a pair or mob session with an agent and defined roles: driver, navigators and the agent as typist.
  • Rotate roles on a timer so everyone practises prompting, reviewing and steering.
  • Make prompting and review decisions visible to the whole group so tacit knowledge spreads.
  • Recognise when the group should stop the agent, take over manually or start again.
  • Decide when ensemble work with an agent pays off and when solo or async work fits better.

Agenda

  1. Module 1

    New roles for a proven practice

    • Driver, navigator, typist: what changes when the agent writes the code
    • The person at the keyboard prompts; the group thinks ahead and reviews
    • Working agreements: when to interrupt, how to disagree
  2. Module 2

    First rounds

    • Exercise: pair session with Claude Code, Cursor or GitHub Copilot on a small task
    • Exercise: mob session on a real ticket from your backlog
    • Timer, rotation and short retros between rounds
  3. Module 3

    Learning from each other

    • Thinking aloud: why this prompt, why reject that diff
    • Capturing good prompts and team rules on the spot
    • Spotting over-trust and drawing in quieter team members
  4. Module 4

    Making it a habit

    • Formats: weekly mob hour, pairing on complex tickets, onboarding sessions
    • Remote ensembles with screen sharing and shared sessions
    • Retrospective: what your team keeps

Why practise agentic coding together

Agentic coding is a skill that is hard to see from outside. Two developers with the same tool get very different results, and the difference lies in small decisions: how a task is framed, which context is given, when a diff is rejected. Working together makes these decisions visible.

Ensemble programming also counters a common risk: that a team stops understanding its own code. When several people steer and review the agent's work, knowledge of the codebase grows across the team instead of shrinking.

Roles in an ensemble with an agent

We adapt the classic mob programming roles to working with an agent:

  • Driver: talks to the agent and accepts or rejects changes
  • Navigators: plan the next step, watch the diff, raise concerns
  • Facilitator: keeps time, rotation and working agreements
  • Agent: writes code, runs tests and explains its approach on request

Low barrier to entry

No prior experience with coding agents is needed. We bring prepared practice tasks and, if you like, use tickets from your own backlog. The workshop runs on-site or remotely with screen sharing, in English or German, led by neooutput AI Trainers and AI Code Engineers. It is designed so you can continue it afterwards as a regular team slot. Because the focus is shared learning, not performance assessment, the format is usually easier to discuss with a works council.

Frequently asked questions

Isn't it inefficient for several people to watch one agent?

For routine tasks, yes. For complex changes, unfamiliar code and learning phases, ensemble sessions save time later through fewer mistakes, faster reviews and shared understanding. We help you pick the right moments.

Does the format work remotely?

Yes. With screen sharing, a shared session or handing over the driver role via Git branches, remote ensembles work well. If your team is distributed, we practise exactly that variant.

Which agent do we need?

Any agent your team is allowed to use, such as Claude Code, Cursor, GitHub Copilot or OpenAI Codex. The format is tool-agnostic.

Workshop

Get in touch

How about a first ensemble session with an agent, together with your team?

Tell us roughly how many people will join and pick dates that work – we will get back to you within 24 hours with a proposal for content, format and schedule.

Prefer to talk first?

15 minutes, free and without obligation – with our AI Trainers and AI Code Engineers.

Book a call for free (opens in new tab) +49 761 519 751 50 (24/7)

Request a training / workshop

Workshop: Pair & Mob Programming with an AI Agent: Team Workshop · Half day (approx. 4 hours)

Dates that work for you (optional)

Suggestions about 2 weeks ahead – pick as many as you like

Reply usually within 24 hours · jocapps is ISO 27001 certified · Privacy