Contact us
Who We Build For
AI Solutions
Start
Build
Run
Before you start
How We Work
Transparent. A team that stays. Your IP: code, trained weights and data are yours by contract.
Read how we work
MusicTech

It-Jim develops music technology for creative production, including music recognition, generation and voice processing.

Go to page
Startups

Software development services for startups. It-Jim helps founders build MVPs, AI solutions, and scalable products from idea to market launch.

Go to page

Not sure where to start?

Talk to an expert

03 · Proof

AI proof of concept on your own data

An AI proof of concept on your data: a model trained on your recordings and a working demo your pilots can use. Converted and timed on an average phone where on-device is the target. Six to twelve weeks, at a price set by the research.

Timeline6 to 12 weeksPriceEstimated from the research
Send what you haveWhat you get
Accuracy before and after training, 42% to 86% on the same held-out test set, and the demo with a rep counter
How we workTransparent. A team that stays. Your IP.
Read how we work
  • You own the IP

    Code, trained weights and your data are yours by contract. We sign an NDA before you send anything.

  • A team that stays

    Knowledge is documented and shared across the team, so nothing depends on one person. You can ask to change anyone on the team, and the work carries on.

  • Transparent work

    The whole team is in your Slack and on your calls. Regular demos and reports, and we tell you about problems as soon as we see them.

Who it is for

Is this the right step for you?

  • Pilots were sold on automatic measurement and the current model fails on real data.
  • You are raising and need a demo with numbers behind it.
  • You are adding a new sport, a new test or a new measurement to a product that already ships.

Proof

Results from past projects

Face with a heat map of the three skin regions the camera reads the pulse from: forehead and both cheeksCase studyVastmindz

Vital Signs from a Selfie Camera

  • A selfie camera app that measures heart rate almost as accurately as a finger pulse oximeter, along with blood oxygen, breathing rate, heart rate variability and stress
  • Sold B2B as an SDK to insurers, telemedicine providers and corporate wellness programmes
  • Built to handle millions of users
Computer VisionHealth TechRead the case
Postround golf scorecard recognition appCase studyPostround

AI-Powered Mobile App for Automatic Golf Scorecard Recognition

93% scorecard detection accuracy

What you get

Proof of concept: the deliverables

  1. 01

    A model trained on your data.

  2. 02

    A working demo, as a web tool or a test app, that your pilot users can try.

  3. 03

    An accuracy report, before and after, on a test set fixed before training started.

  4. 04

    Where the device is the target: the model converted and timed on an average phone.

  5. 05

    Code and trained weights. All IP is yours.

  6. 06

    A one-page technical memo you can put in an investor deck.

  7. 07

    A roadmap and a price estimate for the first product version.

How it works

From first message to done

  1. 1

    Weeks 1 to 2: labelling guidelines, labelling, and a held-out test set that nobody trains on.

  2. 2

    Weeks 3 to 8: training and iteration against the failure catalogue from the research. A demo call every week.

  3. 3

    Final weeks: conversion to the device, the demo build, the report.

Timeline6 to 12 weeks
PriceEstimated from the research

The team is a tech lead, one or two engineers and a project manager, all in-house.

Done when

A demo your pilots can use and an accuracy report on your test set that shows where the model works and where it does not yet.

What comes next

When the demo works and you need it in users' hands for the round or the first paying pilots, the next step is the First product version.

04First product version

Compared with the alternative

An SDK demo shows you the vendor's data. A PoC shows you yours, with a test set you can audit and a number you can quote. If the number is bad, you have learned it for a fixed price, with the failure analysis and the code in hand.

Questions

What teams ask about proof of concept

Some ideas need more than one round. You keep the code, the weights and the failure analysis, and we suggest what would close the gap: more data, a narrower scope or a different approach.

It stays yours and is used only for this project. We delete it on request when the work ends.

It is a demo build. The first product version adds the capture flow, the screens and the backend that users need.

How we work

What you can count on

Who owns the model?

You do. Model, weights and code are yours by default, and the contract says so.

Will it go over budget and timeline?

The scope is fixed and each milestone has a defined deliverable. If something slips, you hear before the deadline.

How do we judge the result without an ML engineer?

The acceptance test is a number on a test set you can audit.

Next step

Send what you have. Verdict in a day.

Start with the free assessment. The research sets the PoC price. If you already have a research report or a working baseline, book a scoping call.

  1. 1
    You send a spec and a few recordings

    A paragraph on the feature, a link to clips or sensor logs, the platforms. NDA first if you need one.

  2. 2
    An engineer reviews it

    The CTO or a domain expert goes through the measurement, the conditions and the platforms.

  3. 3
    You get the verdict within a day

    Whether the feature can be measured, the risks we see, and the recommended next step, in writing.