Kaidris AI products & production AI
Kaidris builds AI products — gooolll, Kaidris ABAP, Blocks, Redline Legends, AI Bro and open-weight models — and brings the same practice to companies taking AI to production. Working software, not slideware.
01 Products
Every product here was built with Claude Code, Anthropic's coding agent — for Kaidris ABAP, the training and evaluation tooling — and gooolll and AI Bro also call the Claude API while they run. Each one is listed with its real status: live, released, or in development.
Scores, fixtures, standings and stats across the major competitions, in several languages, with match write-ups and analysis drafted by Claude. Built in eleven weeks by a crew of seven coordinated AI agents and one person.
A free 7B model that writes modern ABAP offline on one GPU: 51 of 158 general tasks, against 22 for its base model. Forge, with a compile-test-fix agent, is in development.
A calm puzzle with a daily challenge, three board sizes and 21 badges. Fully offline; no data collected.
Offline-first drag and circuit racing with a career, upgrades and AI opponents. Not released yet; the source is public.
Message it on Telegram. It answers itself, hands the job to one of your own machines, or calls a stronger model — inside a hard monthly budget it cannot exceed.
Frontier-scale builds that run on one consumer GPU, plus an Arabic fine-tune. 20,000+ downloads, MIT and Apache-2.0.
02 What we do
We help organizations adopt AI that actually reaches production — pairing hands-on AI engineering with 15+ years of enterprise depth.
Most AI initiatives stall between the pilot and the platform. We work the whole path: from the business problem, through architecture and build, to a system running inside your processes with governance that holds.
The firm is led by a practitioner who still builds — so the advice is grounded in what ships, and the engineering is grounded in how large organizations really work.
03 Services
Turn business problems into feasible, high-value AI use cases — scoped, prioritised and tied to measurable outcomes before a line of code is written.
Architecture, orchestration and guardrails for systems of agents that hold up in production — observability, failure handling and reliability first.
Fine-tuning, quantization and deployment — including cost-efficient single-GPU inference and open-weight models where they make sense.
Quality controls, red-teaming and “prove-it-can-fail” verification frameworks — so confidence in a system is earned with evidence, not assumed.
Wiring AI into SAP and your core business processes and data — where the value lives, with the controls enterprise systems demand.
A senior AI architect for teams that don't have one in-house — direction, reviews and hands-on help, a few hours a week.
04 Why Kaidris
A small firm, deliberately. You work directly with the principal — the same person who designs the system, builds it and stands behind it in production.
Public open-weight models with 20,000+ downloads and a production multi-agent system — the same hands that advise, build. No hand-off between the people who talk and the people who deliver.
Evidence: public work, running systems15+ years across SAP, digital transformation and large-scale delivery. We know how big organizations change — and where AI projects really fail.
Evidence: 15+ years enterprise deliveryEnglish, Arabic and French — across the GCC, Europe and global teams. Workshops, documentation and stakeholder alignment in the language the room speaks.
Evidence: EN · AR · FRBefore we start, we write down what "done" looks like and how it will be measured in production — then every engagement is held to that, not to a deliverable that lives in a slide.
Evidence: measured, not asserted
05 Proof
AI transformation leader and enterprise architect. Fifteen-plus years in enterprise software (SAP) and digital transformation; now builds and ships production AI — multi-agent orchestration, open-weight model builds and AI governance.
06 How we work
A defined scope with a defined outcome — a use-case portfolio, a target architecture, or a system taken to production.
Ongoing access for the decisions that shouldn't wait — architecture reviews, model and vendor choices, governance questions.
A senior AI architect embedded a few hours a week — for teams that need the seniority without the headcount.
Remote-first, worldwide — with on-site time when it matters. Every engagement is scoped to the problem in front of us: there is no menu, there is a conversation.
Remote-first · Worldwide · EN / AR / FR07 Work with us
Tell us about the problem. We'll tell you plainly whether AI is the right answer — and what it would take to put it into production.
