One way of working with agents, for the whole team.
One day with up to 8 of your developers, in your repository, on one tool: Claude Code, Codex or Cursor. The team ends it with the agent config, a task template, test expectations and a review checklist in the repo, and a follow-up call within 30 days.
15 minutes on video with Michał Jaskólski. Afterwards you get a written proposal, or he tells you it is not a fit.
Results that depend on who opened the PR.
Your developers already use coding agents. Each one set theirs up alone, so each gets different results. The review queue absorbs the difference.
-
Two developers, two setups.
One has a project config and starts from a spec. The other pastes the ticket into the chat. Their pull requests show it.
-
Review becomes the bottleneck.
The agent writes faster than people read. Whoever reviews checks every line, because nothing records what the agent was asked to do.
-
More changes, more incidents.
When code arrives faster and nobody has agreed what must pass before review, more fixes come back from production.
-
Nothing to show.
When the board asks what the agents changed, you have anecdotes.
Faros AI measured both sides across 22,000 developers (AI Engineering Report 2026 (opens in a new tab), April 2026): tasks completed per developer +33%, incidents per pull request +242%, median time a change waits in review +441%. Vendor telemetry.
What happens on the Team working day.
You work with Michał on your own code and on tasks from your own backlog. What you set up stays in the repository, where the agent reads it on every run.
Before
Preparation
- A preparation call with your tech lead to pick the repository, the tool and two or three real tasks from the backlog.
- Michał reads the repository's structure, build and tests, with read access you grant for that week, or on a shared screen if the code may not leave your network.
- Everyone attending has the tool installed and signed in under your own licences.
On the day
About 5 hours live
- Write the agent config from the conventions the repository really follows.
- Turn a backlog task into a brief, using the task template.
- Hand the real tasks to the agent, working in pairs.
- Agree the test expectations: what has to pass before anyone reviews.
- Review the agent's pull requests against the checklist, and fix the checklist where it missed something.
Within 30 days
Follow-up
- Your developers commit the four files during the day, on a branch the team reviews and merges.
- A 60-minute call within 30 days: what held up in daily work, what did not, and what to change in the files.
What stays in your repository
The next time anyone on the team starts a task, their agent reads the same config, the brief follows the same template, and review checks the same list.
CLAUDE.md,AGENTS.mdor.cursor/rules/- Agent config: the conventions, commands and limits of this repository, in the file your tool reads.
- Task template
- The brief every task starts from: what to change, what done means, what not to touch.
- Test expectations
- What has to pass before a change goes to review, and which tests a change must add.
- Review checklist
- What a reviewer checks in an agent's pull request, in order.
Not included
- Production deployment
- Security certification or a compliance sign-off
- Support beyond the follow-up call, which is what Ongoing care is for
- Tool licences: your team uses its own Claude Code, Codex or Cursor plans
- A second repository or a second tool
- On-site visits: the day runs remotely
The day includes a Team Toolkit ($299.99 on its own): every guide, all seven books and the full Setup Pack generator for 12 months, for up to 10 people on your team.
First two teams: $1,999 instead of $2,499. Why
Smaller steps and bigger ones.
Each service has a fixed price and a written scope cap. Work past the cap is quoted separately, before it starts.
-
about 3 h
Repo diagnostic
When you want an outside read before committing the team's time.
One repository, about 3 hours of Michał's work, written findings and a recommended next step.
The diagnostic fee is deducted from a Team working day or an Audit + setup booked within 60 days after it.
$499
-
half a day
Half-day workshop
When the team needs the same basics before working in its own code.
Up to 10 participants, the standard curriculum, remote.
- The autonomy ladder and where the team stands on it
- Briefing an agent: intent, spec, plan
- The config files an agent reads
- Tests as the stop condition
- Reviewing an agent's pull request
$1,499
-
1 day
Team working day Start here
When you want the whole team working with the agent by the same rules after one day, in its own repository.
Up to 8 people, one repository, one tool, about 5 hours live, a 60-minute follow-up within 30 days, and a Team Toolkit.
What happens on the day Ask about this service (opens your email)
$2,499
-
up to 4 days
Audit + setup
When the setup has to cover a whole repository and its toolchain, which takes more than one day.
2 weeks, at most 4 working days of Michał's time, one repository and its toolchain, acceptance criteria agreed in writing before work starts.
$4,999
-
up to 10 days
Pilot implementation
When one team should run the method for six weeks, with someone accountable for it.
6 weeks, at most 10 working days of Michał's time, team and repository limits written into the proposal.
from $12,999
-
up to 4 h a month
Ongoing care
After an implementation, while the tools keep changing.
Only after Audit + setup or a Pilot implementation. Up to 4 hours a month, the supported configurations named in the agreement, a response window stated in it.
$999 a month
Prices in USD, excluding VAT. Invoiced by Wondel.ai sp. z o.o.
Only need the guides? Team Toolkit gives up to 10 named people 12 months of every guide, all seven books and the full Setup Pack generator, for $299.99 on one company invoice. Bought online, no call.
From the call to the date.
Each service starts with a call and a written proposal. Nothing is bought online.
-
Book a 15-minute call.
Free, on video. Tell Michał about the team, the tool and the repository. If no service fits, he says so on the call. You can take the Tech Lead Scorecard first and bring the result.
-
Get a written proposal.
Scope, dates, price and what is excluded, in one document.
-
Accept it and fix the date.
The proposal suggests dates. The one you pick is held once you accept.
-
Pay within 14 days of the invoice.
Wondel.ai sp. z o.o., Warsaw, invoices after delivery for the Repo diagnostic and the Team working day, and on the schedule in the proposal for the rest.
Prefer to write first? Email teams@developertoolkit.ai with the service in the subject, how many developers, which tool and where the code lives.
Who it is for.
- Teams of 4 to 30 developers who already use Claude Code, Codex or Cursor, each in their own way.
- A tech lead, engineering manager or CTO who can choose one repository and one tool for the day.
- Teams ready to change how they brief and review agent work, starting with one repository.
Who it is not for.
- A team that has never worked with a coding agent. The half-day workshop is the better start.
- Anyone who needs a productivity percentage promised in writing. These services do not promise one.
- Production deployment, security certification or a compliance sign-off.
- Work done for the team while the developers are elsewhere. The working day needs them in it.
Michał Jaskólski runs every engagement himself.
He answers for every guide on this site, in English and Polish, and builds with coding agents every day as CTO at TasteRay.
- Years shipping software
- 26
- IPOs
- 2
- CTO at
- TasteRay
- AI lecturer at
- SWPS University
- Created wondel.ai Skills
- 2.3k+ GitHub stars (opens in a new tab) read October 10, 2026
What you can check before the call.
Questions before you book.
Anything else? Ask on the call, or by email.
Is the working day remote?
Yes. It runs on your own video tool: your developers work on their own machines and Michał follows on a shared screen. The half-day workshop is remote as well. Michał runs every session in English or Polish.
Which tool should we pick?
The one your team already pays for and uses most. The day covers one: Claude Code, Codex or Cursor. If the team is split, pick the tool for the repository you bring; the task template, test expectations and review checklist work with the other two as well. The comparisons set all three side by side.
What access to our code do you need?
For the day itself, none: your developers drive on their own machines and accounts, and Michał follows on a shared screen. For the preparation, read access to the one repository for that week, or a screen share if the code may not leave your network. The agent runs under your own Claude Code, Codex or Cursor plans, so your existing data terms with those vendors apply.
Will you sign our NDA?
Yes. Send it before the preparation call, and it is signed before you share any code.
What does our team need to prepare?
A repository the agent can build and test locally, the tool chosen in advance with a licence for everyone attending, and laptops with the tool installed and signed in. The preparation call goes through the list, so the day does not start with installing things.
We have more than 8 developers. What then?
The working day is capped at 8 so everyone gets time at the keyboard. A larger team can book a second day for a second group, or start everyone on the half-day workshop, for up to 10 people. For several teams at once, the pilot implementation sets the team and repository limits in its proposal.
How is it invoiced?
Wondel.ai sp. z o.o., Warsaw (NIP 5252951298), invoices after delivery for the Repo diagnostic and the Team working day, and on the schedule in the proposal for every other service. Prices exclude VAT: a company in Poland pays Polish VAT on top, and for a company elsewhere the proposal states the VAT treatment. You pay by bank transfer within 14 days of the invoice date.
Can it be paid from public training funds?
Wondel.ai is not registered with any public training-subsidy scheme, so no co-funding is promised here.
What if it does not work for our team?
You find out in one day, at a fixed price, and the invoice comes after the day. The files are in your repository either way. The follow-up call within 30 days is for what did not hold up in daily work, and the files change to match. No service here promises a productivity percentage, because nobody can measure one honestly in a day.
Why do the first two teams pay less?
The Team working day is a new service. The first two teams pay $1,999 instead of $2,499, and in return fill in two short written feedback forms, one after the day and one after the follow-up call. Being quoted by name is your choice and not a condition. Once two teams have booked, the price is $2,499 and this offer comes off the page.
Start with 15 minutes.
Tell Michał about your team, your tool and your repository. You get a written proposal with scope, dates and price, or a straight answer that none of this fits yet. Without a shared method, every new hire sets the agent up their own way and review keeps catching the difference.
The scorecard is free and takes about 8 minutes. Prices exclude VAT. The Repo diagnostic and the Team working day are invoiced after delivery.