Add AI features to your app that work outside the demo.
I build AI features into existing apps, and agents that do repetitive work for your team. I built Simor’s AI meal logging, which turns a photo, a voice note, a barcode or a line of text into a logged meal, and AI agents build my own projects.
From $10,000
One fixed price for the agreed feature. The model provider bills its usage to your own account.
Request an AI proposal- Who it’s for
-
Businesses with a working app or internal tool that want AI to do one specific job: read documents, answer customers, sort requests or turn photos and voice into data. You know the job; I make the AI do it and measure how often it gets it right.
- What you get
-
One AI feature or agent, live in your product: the model setup and prompts, the code around them, a test set that scores the answers and an estimate of what each use costs.
- Timeline
-
Usually 3 to 6 weeks from the signed scope to the feature live in your app.
- Included
-
- Choosing the model and provider for the job, with the trade-offs in writing
- The feature built into your app, with the screens it needs
- A test set of your real examples that scores the AI’s answers on every change
- Limits, fallbacks for when the model fails and logs you can read
- A cost-per-use estimate before launch
- 12 months of free fixes for bugs in the work I delivered
- Not included
-
- Training a model from scratch
- Model usage fees, which go on your own provider account
- Collecting or labelling large datasets
- A promise that the AI is never wrong; the test set shows how often it is right
- How to start
-
Send the request form with the job you want the AI to do and a few real examples. I reply with what is realistic, a written scope and one fixed price.
What this can look like.
-
Input by photo or voice
People snap, say or scan instead of filling in forms. Simor logs a meal from a photo, a voice note, a barcode or typed text.
-
Answers from your own documents
A support assistant that answers from your help centre and hands over to a person when it isn’t sure.
-
Agents for back-office work
An agent that reads incoming requests, sorts them and drafts replies for your team to approve.
I build with AI agents myself.
My own projects are split into small tickets. Agents pick them up, each in its own workspace, build them test-first and open a pull request with screenshots and a video. I review every one before it ships.
Questions.
How much does it cost to add AI to an app?
With me, from $10,000 for one feature or agent, at a fixed price. On top of that the model provider charges per use, and you get a cost-per-use estimate before launch.
Can I integrate ChatGPT in my app?
Yes, through the provider’s API. Most of the work is everything around that call: what the model sees, what it may do, what happens when it is wrong and what it costs. That is the part I build and test.
What is the difference between an AI feature and an agent?
A feature answers when a user asks, like turning a photo into a logged meal. An agent works through steps on its own, like reading an inbox, sorting requests and drafting replies, and asks a person before anything important goes out.
Will the AI make mistakes?
Sometimes, yes. I build a test set from your real examples so you know how often it is right before launch, and I design the feature so a person can check or undo what matters.
Who owns the code and prompts?
You do. At handover you get the source code, the prompts, the test set and the docs.