Credit is not optional
Every listing names the models, the control methods and the human editing. Undocumented output helps nobody and ages badly.
- Home
- About
About Connecting Ai App
A curated publishing platform for everything people make with artificial intelligence.
Our purpose
We built the place we needed ourselves
Connecting Ai App began with a frustration that anyone working with generative models will recognise. The work was good. The presentation was terrible. Finished projects lived in cloud folders, chat threads and half finished portfolio sites, and every time someone asked to see them the answer was a zip file and an apology.
So we built a platform with one job: take anything created with artificial intelligence and give it a page worth sharing. A page that loads in under a second, shows the work at full quality, credits the models and tools honestly, and tells a visitor in ten seconds what they are looking at.
What we mean by AI work
We deliberately kept the definition wide. A diffusion series belongs here. So does a retrieval assistant, a fine tuned vision checkpoint, a synthetic narration library, a video pipeline, a dataset, and the prompt kit that made all of it repeatable. If artificial intelligence did real work in producing it, it has a home on this platform.
How we think about credit
Every listing carries a tools field, and we treat it as essential rather than optional. Naming the base model, the control method and the human editing that followed is not a confession, it is the interesting part. It is also what lets the next person build on your result instead of guessing at it.
How we think about consent
Synthetic voice and likeness are where this industry earns or loses trust. Audio listings carry a consent field, and a listing that cannot show a release for a cloned voice does not stay published. The same applies to datasets: provenance and licence sit next to the download, not three clicks away.
How the platform is built
The front end is driven entirely from the admin panel. Fonts come from Google Fonts and are switchable per role, the type scale and colour system are numeric inputs rather than hard coded values, and the header, menu style, banners and footer are all composable. That decision was selfish: we wanted to restyle the whole site during a launch week without a deployment, and now everybody gets that.
Good AI work is not a lucky prompt. It is a pipeline somebody documented, tested and then had the confidence to show in public.
Where we are going
Next on the roadmap: collections that let a studio group a body of work under one cover, richer evaluation tables for model pages, and an API so a pipeline can publish a finished render the moment it leaves the render queue.
Six listing typesImage, app, video, audio, model, tool
0+
AI projects published
0+
Creators and studios
0
Countries reached
0%
Platform uptime
How we work
Four commitments we do not negotiate on
These are the rules the platform is built around, and the reasons behind most of the product decisions we make.
Consent before cleverness
Cloned voices and real likenesses need a documented release. No release, no listing, no exceptions.
Speed is a feature
Images are optimised on upload and pages ship without heavy frameworks. A gallery should open before the visitor loses interest.
You own the look
Fonts, colours, header, menu style, footer and every home section are editable. We will not ask you to open a template file.
The people
A small team with strong opinions
Four of us, split across product, engineering, design and research.
Layla Haddad
Founder and Product
Spent nine years building publishing tools for photographers before turning the same attention to AI work.
Marcus Reid
Head of Engineering
Backend and inference infrastructure. Believes a fast page is a feature and a slow one is a bug.
Priya Nair
Design Lead
Type, grid and colour. Designs the templates that make an upload look like a considered release.
Tomas Berg
AI Research
Evaluation, safety and model documentation. Writes the checklists nobody enjoys and everybody needs.
Questions worth answering
Everything people ask before their first upload
Licensing, consent, file limits and how much of the design you control.
Anything made with or substantially assisted by artificial intelligence: still images, applications and agents, video, audio and music, fine tuned models, datasets, prompt kits and automation workflows. If AI played a real part in producing it, it belongs on the platform.
Yes, and it works in your favour. Every listing has a tools field and an optional prompt or training notes block. Visitors trust documented work far more than mystery output, and other creators can build on what you share.
You do. Publishing a listing grants the platform permission to display it and nothing more. You set a licence on each listing so visitors know what they may and may not do with it.
Audio listings include a consent field. If a listing uses a cloned voice or a real person likeness, you record the release covering it. Listings that cannot show consent are removed.
The whole front end is driven from the admin panel: Google fonts, type scale, colours and gradients, header layout and menu style, logo size, banners, footer columns and every section on the home page. No template editing required.
Individual images are capped at eight megabytes and are downscaled automatically for delivery. For video and large model files you link to your own hosting or object storage and the listing embeds it.
Every listing is indexed by title, description, tools and tags, and the showcase can be filtered by category and by kind. Tag pages let a visitor follow a thread from one project to related ones.
New listings from new accounts are reviewed before they appear publicly. We check that the work is real, that the credits make sense and that nothing violates the consent and licensing rules.
Publish your first project this week
Bring an image series, an assistant, a model card or a sound library. We will make it look like a release.