Case studies · Field notes
How we operate — and what we learned building it.
Not testimonials. Working notes from real, large-scale systems we shipped — written to transfer know-how. Each study lays out the problem, the approach, the outcome, and a takeaway you can reuse.
Case 01Mobile · Cross-media
Putting mobile-to-IT on the map: a 1,500-app Java catalogue at 100M+ downloads
Before app stores existed, distribution ran through carrier decks. We industrialised the pipeline — authoring, porting and QA of Java (J2ME) apps across hundreds of handset profiles — and pushed a 1,500+ title catalogue through Indian mobile network operators, reaching over 100 million downloads.
Problem
- Fragmented handsets, dozens of screen and API profiles
- No unified store; carrier decks were the only channel
- Manual porting did not scale to thousands of SKUs
Approach
- A build matrix that auto-targeted device profiles
- Automated QA against real handset farms
- Revenue-share integration directly into operator billing
Outcome
- 1,500+ titles live across operator decks
- 100M+ cumulative downloads
- A repeatable porting pipeline, not one-off builds
1,500+
Java titles shipped
100s
Handset profiles supported
TakeawayStandardise the pipeline before you scale the catalogue. The moat was never a single app — it was the repeatable system that turned device fragmentation into a solved problem. We apply the same principle to AI delivery today.
Case 02Broadcast · Telco
TV-SMS overlay broadcast over Galaxy 25: turning local TV into a two-way channel
Through a partnership with Blue Frog Mobile (Seattle) and satellite distribution over Galaxy 25, we delivered an on-air SMS overlay system to broadcasters — working with networks including Phoenix TV while concentrating on local TV stations. Viewer messaging was monetised on a revenue-share basis connected to AT&T, Verizon and T-Mobile.
Problem
- Broadcast was one-way; no viewer interaction loop
- Local stations lacked interactive monetisation
- SMS billing spanned multiple US carriers
Approach
- Real-time SMS overlay rendered into the broadcast feed
- Satellite delivery via Galaxy 25 to affiliate stations
- Carrier revenue-share settlement across AT&T, Verizon, T-Mobile
Outcome
- Interactive on-air segments for local broadcasters
- New per-message revenue line for stations
- A model proven with networks such as Phoenix TV
Galaxy 25
Distribution satellite
Local-first
Station strategy
TakeawayThe value sat in the least glamorous layer: clean settlement across carriers. Whoever owns the boring plumbing owns the business. In AI, that plumbing is data flow, evaluation and deployment — which is exactly where we focus.
Case 03Governance · Method
Operationalising AI ethics: turning five principles into a shippable checklist
An ethical framework only matters if it survives contact with a deadline. We translated ImageFirm's five principles — transparency, agency, growth, responsibility, sovereignty — into concrete, testable gates inside the 24-hour delivery pipeline, so that ethics is enforced by the process rather than left to good intentions.
Problem
- “AI ethics” is usually a PDF nobody applies under pressure
- High-stakes automations can quietly remove human judgement
- Buyers can't verify claims about data use or lock-in
Approach
- Each principle mapped to a pass/fail gate in delivery
- Mandatory AI-use disclosure and human-override points
- Source-code handover and no-black-box rule enforced by default
Outcome
- Ethics checked on every build, not just described
- Clear audit trail of where humans stay in control
- Trust made verifiable, not merely asserted
5
Principles as delivery gates
100%
Builds ethics-reviewed
TakeawayValues that aren't wired into the workflow don't survive deadlines. Encoding ethics as pass/fail gates is what makes “human-first AI” a practice instead of a slogan — and it is fully reusable by any team.
Case 04Method · AI today
The 24-hour delivery method: how a working AI solution ships in a day
Our current practice compresses discovery-to-deployment into a fixed 24-hour clock (24–72h for platforms) without cutting corners — by pre-committing scope, reusing hardened components, and testing in the open before any payment.
Problem
- Traditional cycles hide risk in months of estimation
- Buyers pay before they see anything working
- Scope creep erodes both trust and margin
Approach
- Fixed scope and price agreed within the first hour
- A library of hardened, reusable AI components
- Private test environment before any commitment
Outcome
- A working, deployable solution in 24 hours
- Source code and 12-month updates handed over
- Payment only after the buyer validates the result
100%
Source code ownership
TakeawaySpeed is a by-product of constraint, not haste. Pre-deciding scope and reusing proven building blocks is what makes a same-day, ethics-clean delivery repeatable — for you, not just for us.