AI and machine learning

AI built into the work your company already does, and tested on your own cases before anyone relies on it.

What can be built

The task comes first. The model is chosen to suit it, and sometimes the answer is a simpler one.

Language models on your data: Connecting a language model to your own documents and systems, so that it answers from what your company knows.
Text mining: Pulling structure out of reports, reviews, emails and notes that nobody has time to read.
Document processing: Reading invoices, contracts or forms automatically and passing the details to the systems that need them.
Prediction: Models that forecast demand, churn or risk from your own history.
Automation: Software that carries out tasks of several steps, with a person approving the ones that matter.
Evaluation: Testing a model on your real cases before anyone relies on it, and watching it after launch.
The data underneath: Cleaning and connecting the data a model needs, which is usually most of the work.

Past work: AI before ChatGPT

At a Fortune 500 company, Amos joined the strategy and planning department of one division and introduced AI and text analytics there, years before ChatGPT. He built natural language processing and text mining tools that drew signal out of the company's unstructured data, and marketing, R&D and strategy all used them.

Later, at OurCrowd, he built AI into the team's everyday work.

“What also stood out was his ability to programmatically integrate AI into our workflows at a time when most people in the industry weren't even thinking about it.”

Micky Sapir, Vice President of Product, OurCrowd

How to start

The brief

Describe the task you want done better or faster, and where its data lives today.

A first version

Work usually begins with something small that runs on your real data, so you can judge it before committing to more.

Who does the work

Amos Weiskopf founded Tevelis. He studied computer science at the University of York and has written software for more than twenty years, along with the data engineering that AI systems depend on.

Questions buyers ask

Do I need a custom AI model?

Often not. Many tasks are handled well by an existing model connected to your own data. A custom model is worth building only when that falls short.

What is the difference between AI and machine learning?

Machine learning is the part of AI that learns patterns from data. Most of what businesses now call AI, language models included, is built on it.

Is our data safe?

Your data stays under your control. Whether any of it may be sent to an outside AI provider is agreed with you before anything is built.

How do you know a model works?

It is tested on a set of your real cases with known answers, before launch and again once it is in use.

How long does an AI project take?

A first working version is often ready in weeks. The full timetable is agreed in the brief.

What does AI development cost?

That depends on the task and the systems it has to work with. The price is agreed after the brief.

Engagements by written brief.

Work begins from a documented problem statement, not a sales call.

Write a brief Or book a meeting