Key takeaways
- Coaching works on your repository and your problems; a group workshop works on what most participants have in common.
- It suits tech leads preparing a rollout, solo developers and founders shipping an app, and experienced engineers who are sceptical of agents.
- A series of short sessions with real work in between beats one long session.
- Coaching a few champions first is often the most efficient way to prepare a whole team.
Who gets the most out of coaching
Group workshops are efficient for teams that share a codebase, a tool and a starting level. Coaching fits when one of those is missing – or when one person's progress matters disproportionately.
- The tech lead before a rollout. Wants to try things on the real codebase first, settle the configuration and the context file, and then teach the team with confidence.
- The solo developer or small-studio founder. Has no colleague to compare notes with and needs answers on their own stack, not a curriculum.
- The sceptical senior engineer. Tried an agent once, got a mess and stopped. Needs to see it work on a hard problem in code they know well.
- The founder with an AI-built app. Built a product with Lovable, Base44 or a coding agent and now has to understand it well enough to maintain it and get it into the stores.
- The engineering manager. Doesn't code daily any more but has to make decisions about tools, budgets and policies and wants hands-on experience first.
Coaching or group workshop?
| 1:1 coaching | Group workshop | |
|---|---|---|
| Content | Your questions, your repository | Shared curriculum, adapted to the team |
| Pace | Set by one person | Set by the group |
| Format | Series of short sessions | Half day to two days |
| Team effects | Indirect, through the coachee | Direct – shared vocabulary and agreement |
| Best for | Leads, champions, solo builders | Whole teams adopting a tool together |
The two combine well. A common pattern is coaching for two or three champions, followed by a team workshop they help run. The overview of what a team workshop covers helps with the comparison.
Talk to an AI Code Engineer about your agentic coding workshop.
In a free call we look at your team, your codebase and your goals and suggest a workshop format that fits – remote or on-site, in English or German.
How a coaching series runs
- Intake call. We clarify goals, the codebase, the tools and what "better" would look like in a month. The goals are concrete: "the agent can add a screen following our architecture without me fixing it afterwards" rather than "get better at prompting".
- Working sessions. Typically sixty to ninety minutes, weekly or every other week, over video with screen sharing. The coachee drives; the coach watches, asks, suggests and occasionally takes over to demonstrate.
- Work in between. Each session ends with one or two tasks for the coming days – a context file change, a ticket to try with a new approach, a hook to configure. The next session starts with what happened.
- Async questions. Short questions between sessions go through a chat channel, so a blocker doesn't wait a week.
- Wrap-up. The last session produces a personal playbook: the configuration, the prompts and workflows that worked, and the open questions for the next stage.
Typical topics, by situation
Legacy codebase
Getting an agent to understand a large, old project: mapping the architecture into the context file, letting the agent write characterisation tests before it changes anything, splitting refactorings into reviewable steps.
Preparing a team rollout
Team settings and permissions, the shared context file, which tasks to hand to agents first, how to measure adoption, how to handle the first incidents. The more advanced material from the advanced Claude Code workshop often comes up here.
Shipping a mobile app
Build loops on Xcode and Gradle, keeping the agent away from signing and store credentials, preparing for review. Where the store release itself is the hurdle, our AI Code Engineers can also take it over – see hiring AI Code Engineers for how that works.
From AI builder to code
For founders who started in an app builder: reading the generated code, moving to GitHub and a coding agent, and deciding what to rebuild. The AI app builder workshop covers the same ground for groups.
How to prepare for the first session
- Have the tool installed and working on a paid plan or with an API key.
- Have the repository open, building and with tests running.
- Bring two or three real tasks: one you expect to be easy, one you have postponed, one where an agent failed before.
- Note what annoys you about working with agents today. Those notes are usually the best agenda.
- Decide which parts of the screen you can share – close windows with customer data or credentials before the call.
What progress looks like after a few sessions
Progress in coaching is easy to see if you look at the work rather than at confidence levels. Good signs after three or four sessions:
- The agent's first plan is right more often, because the context file now says what it used to get wrong.
- Tasks that used to end in a half-finished mess are now split into steps that each end in a passing build.
- The coachee stops an agent run early when it drifts, instead of trying to rescue it with ever longer prompts.
- Pull requests come with a description and test evidence that a colleague can review without asking questions.
- The coachee can explain the setup to a colleague – the clearest sign they are ready to act as a champion.
Your code stays where it is
Coaching runs on your machine, in your accounts, through screen sharing. We don't need a copy of your repository, and we don't ask for credentials. jocapps GmbH, the company behind appsubmitter.io, is ISO 27001 certified, and the same rules apply to coaching as to our customer projects: access only to what the session needs.
Sessions run in English or German. If you are unsure whether coaching or a workshop is the better start, book a free call and describe your situation – sometimes the honest answer is that a single session is enough. The article on workshop costs explains how to weigh the options.
Frequently asked questions
How many coaching sessions are typical?
Most series have four to eight sessions over one to two months. Fewer sessions work for a narrow goal such as setting up a team configuration; longer series suit people who are changing how they work day to day.
Does the coach write code for me?
Occasionally, to demonstrate an approach. The point of coaching is that you can do it yourself afterwards, so you drive the agent and the coach advises. If you need the work done rather than learned, our AI Code Engineers can take it on as a project instead.
Can a non-developer book agentic coding coaching?
Yes – founders and product people who build with AI app builders or coding agents are a common group. The sessions then focus on understanding the generated code, working safely with it and knowing when to bring in developers.
Can coaching be in German?
Yes. Sessions run in English or German, whichever the coachee prefers.
Free, 15 minutes, no obligation
Plan your agentic coding workshop with us.
Our AI Code Engineers run hands-on workshops on Claude Code, OpenAI Codex, Cursor and AI app builders – and ship apps to the App Store and Google Play every week. Tell us about your team.
AI tools, store policies and requirements change quickly – always check the current documentation of the tools and stores you use. This article is independent advice and not affiliated with or endorsed by Apple, Google, Anthropic, OpenAI or any other company or product mentioned; all trademarks belong to their owners.