touchgrass

The Touchgrass charter

Not racing to the bottom.Born there.

Touchgrass makes tools that get the people of Berlin out into their own city. This page says what we believe, how we will behave, and where things honestly stand today.

Draft 1October 2026For discussion, not indexed

If you only read one thing

The whole idea in a minute.

  1. 1Most products get worse over time, because they have to serve you and earn from you at once. In the end the earning wins.
  2. 2Software is now cheap to build. A service for one city, not the world, can afford to get the details right.
  3. 3So we build tools for Berlin where the people and the venues come first and the company comes last.
  4. 4Each tool costs what it costs to run, in the open. Money is an afterthought, and it never decides what you see.
  5. 5We start with one question from one resident: I'm free now, or tonight. What's good?

The charter

What we hold ourselves to.

Every Touchgrass tool is checked against these twelve points. They describe behaviour, not technology.

01

Purpose

Touchgrass gets the people of Berlin out into their own city.

We want a symbiotic relationship with Berlin: the city, the venues and the people.

If you are rewarded and not exploited, why would you not want it?

02

Why this can exist now

Product experiences are deteriorating fast, because they try to serve the user and earn a profit at the same time. Inside every company, the steering eventually turns towards revenue.

Two things have changed. Building software costs very little. And a service that serves only Berlin can afford details a global product never will.

Together they make room for tools that are new, sensible and effortless, and that treat nobody as a second-class citizen.

03

Stance

  • The users and the venues come first. The company comes last.
  • The product experience is king. Without data we are ignorant, so data is queen.
  • We do not exploit the user or the venue in any way.
  • We do not compete unfairly.
  • If someone beats us on both the experience and the price, we welcome it.

04

Money

The aim is to normalise the amount of money exchanged for value. Today a service is a minor value for the user and a major value for the corporation.

  • A tool has to prove its value to the user before it earns anything.
  • Each tool charges what it costs to run, and shows that openly.
  • The cost to run includes a normal salary for the people who work on it, under a published cap.
  • We take commissions where they already exist. They are always disclosed where you see them, and they never decide what we show or in what order.
  • If free is better for you, we push free.
  • Money left over goes towards access for people in Berlin who did not have it.

05

Residents first

Visitors benefit as a side effect.

A visitor hasA fixed place and flexible time.
A resident hasFixed time and a flexible place.

You live here, so you will be back. What matters is what fits the time you have. Touchgrass is built for that case.

06

One job, one slice

Touchgrass is the umbrella. It holds this charter. Under it are slices: separate tools, each defined by the job someone needs done and the experience that job demands.

A job that needs a different experience is a different slice. A slice is not a content category. Treating every job as the same experience is the mistake the big horizontal products make.

Slices can share data and code. They never share an experience by default. Nothing is promoted to shared before it has proved itself in one slice.

07

The first moment

I'm free now, or tonight. What's good?

Planning the weekend, not missing something and discovering what you did not know existed all matter too. They come after.

08

Data

Data arrives in layers.

  1. We find the stable things ourselves.
  2. That foundation earns users.
  3. Users contribute what they see, including the pop-ups no crawler finds.
  4. That draws venues and organisers in to list directly.

Every fact carries where it came from and when. Unknown stays unknown. People who contribute get a better city tool, and credit.

09

What we chase

  • Outings are the goal. People who went from opening Touchgrass to actually going.
  • Reliance is the health check. People who come back week after week.
  • Reach is only a diagnostic.

We never optimise for traffic or for time spent on the site. A tool that gets you out of the door fast is doing its job.

10

What we measure

  • For everyone: a few aggregate, cookieless counts. No identities, no profiles, nothing that follows a person. The numbers are published.
  • For people who choose to log in: full measurement of how they use Touchgrass.
  • Logging in is a choice and everything works without it. An account earns its place in this order: your saved visits follow you across devices, then you can contribute and be credited, then the tool learns what you like.

11

Staying honest

  1. Now: the founder's commitment.
  2. Early: published rules. Open books, cost-based pricing and this charter as public commitments.
  3. Once there is revenue worth protecting: a legal lock, so that nobody can extract profit or sell it. Candidates are a gGmbH, a Genossenschaft or steward ownership.

12

How we work

Humans and AI, together. Work is owned by roles that are responsible for it.

Marketing will run as a pipeline: search first, then social, then paid. Each day the results are reviewed against the goals in point 09, and against nothing else.

The Touchgrass family

Three slices so far.

One is live, one is in development, one is an idea. Each is defined by a job, written the way the person would say it.

Touchgrass / BerlinNo. 01

Culture

Museums and exhibitions.

The job

"When I have free time now or tonight, I want to pick an exhibition I can actually get into, see properly and afford, so I spend that time there and not on finding out."

It began as a personal tool for finding somewhere to go that matched a schedule. Starting from your time is the experience, and it still has to prove itself with other people.

Live

Open Culture
Touchgrass / BerlinNo. 02

RiB

Restaurants, cafés and bars.

The job

Not written yet. It is the next thing to write.

RiB is Restaurants in Berlin: a map of places to eat, with a careful record of where every fact came from. It will be folded into Touchgrass under this charter.

In development, not launched

Touchgrass / BerlinNo. 03

Cinema

Films in your language.

The job

"When I want to see a film, I want a showing in a language I follow, at a cinema where I know what to expect, so nothing about the evening surprises me."

Easy language filtering. Trailers with the right audio and subtitles. What the room and the seats look like, which snacks they sell, what you can bring. Built for every language that lives in Berlin.

Idea

An honest look

Where we stand today.

Two things exist. Both were built fast, mostly by AI agents, before this charter was written. On 1 October 2026 both code bases were read, run and held against the twelve points. This is what came back.

Culture has the experience and no machine to keep its data true. RiB has the machine and no experience. Each is missing what the other has.

Holds meets the charterPartlyNot yetConflicts works against it

Slice No. 01

Culture

A working product people can use today, standing on data that cannot be kept fresh.

109exhibitions at 36 venues, live
4 daysfrom first data to the live site
226automated tests, all passing
12 Octthe day its own freshness rule stops it being republished
  • HoldsBuilt around a job

    It starts from your free time and finds what fits. That is the experience the whole idea began with, and no global product works this way.

  • HoldsFree is pushed, unknown stays unknown

    Free entry windows come first. 18 future ticket prices show as unconfirmed rather than guessed. Every fact links to the page it came from.

  • HoldsMoney never steers

    No commissions, no ads, no paid placement, no tracking. Tickets always link to the venue's own page.

  • PartlyFree now, or tonight

    About 90 exhibitions fit a free afternoon. After 18:30 on an ordinary weekday it is 7 to 12, at 3 to 5 venues. Museums close early, so "tonight" needs the other slices.

  • Not yetData that stays true

    The catalogue is a snapshot from 28 September. Each venue has its own hand-written reader, a changed price at one venue stops the update for all 36, and nothing runs on a schedule.

  • Not yetSymbiotic with venues

    Venues do not know we exist. The one "report an error" link leads to a private page, so nobody can tell us we are wrong. Our reader does not introduce itself to the sites it visits.

  • Not yetOutings as the goal

    The site measures nothing at all, so we cannot tell whether anyone actually went.

  • PartlyPublished rules

    A public page explains sources, coverage and uncertainty. There is no legal notice, no cost page and no charter on the site yet.

The read

Keep the experience, the honesty rules and the design. Rebuild how the data is gathered.

The code underneath was written in a hurry and cannot be maintained as it is. That matters less than it sounds: the valuable part is the idea, and the idea survives a rebuild.

Slice No. 02

RiB

A careful, self-running record of places to eat, with very little built for the person who is hungry.

3,722places on an unlisted test site
845commits in 10 days
85,000dated facts, each with its source
7 : 1lines of pipeline and operator tools for each line of the app people use
  • HoldsEvery fact has a source and a date

    The strongest thing in either project. Facts are never overwritten, can be withdrawn and restored, and nothing is published without a checked dry run.

  • HoldsIt keeps itself up to date

    It has run unattended since 26 September: each night it reads venue sites, checks itself, publishes, and sends a morning report. Five nights published; one run was held back by its own safety check and one started hours late.

  • PartlySomething you can rely on

    All of that runs on one computer at home, with no copy of the data outside that home. The automatic checks on its code have been failing since 25 September.

  • Not yetBuilt around a job

    No user, job or moment is written down anywhere. It began as a map demo. The one journey chosen was "is this place worth it?", and "where do we eat tonight" was explicitly set aside.

  • Not yetFree now, or tonight

    There is no "open now". Opening hours are known for 40% of places but are not part of what the map loads, so they cannot shape the list. The default order is alphabetical.

  • Not yetExperience is king

    Of the 14 specifications written after the first four days, 12 were about data or tools for the operator. The app people see is still the first design. Booking providers and social links are collected and shown to nobody.

  • ConflictsSymbiotic with venues

    Several of the ways it gathers data were chosen for reach, before this charter existed, and would not pass it. Each needs a decision before RiB becomes a Touchgrass slice.

  • Not yetCosts in the open

    Running it is cheap, but there is no ledger and nothing is published.

  • PartlyUseful to other slices

    A core of roughly 10,000 lines is general: reading pages politely in turn, storing dated facts, the nightly run, the publish check. The rest is specific to restaurants. It has no idea of something that happens at a place for a limited time, which is exactly what an exhibition or a film showing is.

The read

The worry was that it was built the wrong way. The finding is narrower: much of it is built well, in the wrong order.

Ten days of work produced a deep, well-evidenced record of four postcodes and the tools to supervise it, and almost nothing that changes what a person looking for dinner sees. The way it stores facts is worth carrying forward. The breadth around it grew ahead of any proven need.

A proposal, not a plan

What comes next.

Nothing here is being built yet. This is the order that follows from the charter and from what the look above found.

  1. Agree the charter.That is what this page is for. Argue with it.
  2. Write the job for every slice before any more code.RiB first. If the job is "where do we eat tonight", most of what it needs is already collected and simply not used.
  3. Write down how we treat venues.We say who we are when we read a site. Every venue can correct us, and can ask to be left out. No method that we would not explain to the venue's face.
  4. Rebuild how Culture gets its data, on the part of RiB that works.Dated, sourced facts that can be withdrawn, plus the thing it lacks: something that happens at a place for a limited time. Written down first, tested first.
  5. Count outings, then put it in front of people."Start from my time" has only ever been tested on its author.
  6. Publish the rules.This charter, the running costs and the salary cap, on the site.

Ideas for later

Not commitments.

  • A paid slice built for visitors.It helps fund free access for residents.
  • Surplus food to the people who need it.Once we understand how restaurants work.
  • Delivery that pays real wages.
  • What only a Berlin tool would know.Pop-ups, construction, a closed street on the way there.

Open questions

Not decided.

  • What is the job of RiB?
  • Which legal form locks the charter in, and when?
  • What can a logged-in person see and delete?
  • Do we buy from Berlin businesses first, as a rule?
  • Where does a normal salary stop?