Products
We create and operate production software under Celikkanat AI. LLMCap, Selltrix and the AI Recommendation Index are the first three products; future products will be developed under the same company as new, validated problems emerge.
SOFTWARE / AI / DATA / AUTOMATION
We build AI systems
for difficult problems.
Celikkanat AI Technologies builds software, intelligent systems and decision infrastructure for companies operating in complex environments.
01 / WHAT WE BUILD
Technology with a clear job to do.
We focus on products that address concrete operational problems. The goal is not to add AI everywhere. The goal is to build software that is useful, controlled and understandable.
02 / HOW THE COMPANY WORKS
Build. Apply. Advise.
Celikkanat AI Technologies operates across three connected lines: we build our own software products, apply AI to difficult operational problems, and advise companies where technology, market decisions and execution meet.
AI Systems
We turn business problems into working AI and agent systems, decision intelligence, data and machine-learning systems, process automation, governance and cost controls, and industrial or enterprise AI.
AI & Digital Commerce
We help companies prioritize products and markets, plan marketplace entry and expansion, build analytics and KPI systems, and run AI-supported commerce decisions across Amazon, Walmart, Shopify, B2B and distributor channels.
03 / PRODUCTS
01
Open LLMCap ↗
LLMCap
Monitor. Alert. Block.
LLMCap is a pre-request LLM spend enforcement gateway. It sits between an application and its model providers, applies dollar-denominated caps, and can block a request before tokens are purchased while keeping cost and policy records in one control layer.
01Real-time spend & request visibility
02Provider, model, token, cost & latency logs
03Pre-request hard caps with Alert and Block actions
04Guided provider setup and integration examples
Actual LLMCap product interface — click any screen to view full size.
02
Open Selltrix ↗
Selltrix
Evaluate. Decide. Move.
Selltrix is an explainable Amazon product-research decision engine. It separates ASIN readiness from niche openness, then combines demand, competition, pricing economics, reviews, offers and evidence quality into a verdict, risk view and action plan.
01ASIN & niche opportunity evaluation from live market signals
02Demand, competition and pricing economics scoring
03Review, offer and evidence-quality analysis
04ENTER / WATCH / EARLY STAGE / AVOID / INSUFFICIENT DATA decisions
Actual Selltrix product interface — click any screen to view full size.
03
View the weekly index ↗
AI Recommendation Index
Measure. Publish. Repeat.
The AI Recommendation Index measures which brands ChatGPT and Gemini recommend to US shoppers. The same buying questions go to both assistants every week; the answers are counted, compared week over week, and published. Brands subscribe to a weekly report: where they appear, where they don't, how the two assistants differ, and which pages the answers rely on. We measure. We don't do implementation — that keeps the index neutral.
01Fixed question sets, repeated weekly on ChatGPT and Gemini
02Brand and model counts, split by assistant
03The sources each answer cites, theme by theme
04Repeatability test: a same-day rerun on the current models moved air fryer brand rates by 3.4 points on average (9 points max)
05Custom question sets and source-gap reports for subscribed brands
Actual AI Recommendation Index page — Week 4, October 2026. Click any screen to view full size.
04
Products in development
IN DEVELOPMENT
AIOptLab
Evidence-driven AI optimization and verification platform.
Digitwiner
Enterprise decision governance and simulation platform.
Underway
Intelligent network for urgent and scheduled local services.
04 / SERVICES
Where we can help.
Outside of our own products, we take on focused engagements where a team needs an AI system designed and built the right way, not bolted onto an existing product.
AI Systems
Teams often add AI to a product without a plan for control, cost or failure handling. We design production AI and agent systems with clear boundaries, human or rule-based intervention, workflow automation, governance and cost control.
Decision Intelligence
Most teams have data but still decide by gut feeling because that data was never turned into a clear recommendation. We turn raw signals into scored, explainable decisions, so a team sees not just a number but why it matters and what to do next.
AI & Data Engineering
An AI system is only as good as the data pipeline underneath it, and brittle pipelines make good models unreliable. We build the ingestion, processing and storage layer a product actually depends on: monitored, tested and built to scale with real usage.
AI-Powered Digital Commerce
We combine product prioritization, market-entry strategy, marketplace and channel expansion, analytics, KPIs and AI-supported operating decisions across Amazon, Walmart, Shopify, B2B and distributor networks.
AI Visibility Measurement
When customers ask ChatGPT or Gemini what to buy, some brands are in the answer and most are not. We track what the assistants recommend in your category, every week, with a fixed and repeatable method — measurement only, kept separate from implementation so the numbers stay neutral.
Operational Digital Twins
A working model of one operational flow: vehicles, orders, a production line, a service counter, or the zones of a water or energy network. We build it from your own records and test it against a past period you know. Only then is it used to compare a change before money is spent, with the assumptions and the uncertainty shown. Where records are thin, we start with a short measurement plan.
FORWARD-DEPLOYED ENGINEERING
Start with one operational problem.
We work close to the people, data and workflow behind the problem. An AI Opportunity Assessment frames what is happening now, what can be measured, and whether a focused pilot is worth building. The problem may sit anywhere in the operation; evidence and practical value decide where we start.
BRINGOne operational bottleneck
WE DEFINEBaseline, value and feasibility
IF IT IS WORTHWHILEScope a focused pilot
05 / APPROACH
Less hype.
More function.
Our product philosophy is simple: use AI when it creates a measurable advantage, and keep the surrounding system clear enough to trust and operate.
Problem first
Start with a real operational pain point, not a technology demo.
Control matters
AI systems should remain observable, governable and understandable.
Build for use
Products should fit into real workflows and produce practical outcomes.
06 / PROCESS
From problem to production.
The same five steps run through everything we build. An opportunity assessment begins in Discover; a focused pilot moves through Design and Build; production work continues through Deploy and Improve.
01
Discover
Understand the operational problem, the data available, and what success actually looks like for the team.
02
Design
Define the system boundaries: what the model should decide, what a person should decide, and how a failure is handled.
03
Build
Build the product itself: data pipelines, models, interfaces, and the controls around them.
04
Deploy
Ship into the real workflow, not a demo environment, with monitoring in place from day one.
05
Improve
Track how the system performs against the problem it was built for, and refine it using real usage data.
07 / PROOF
Built by practitioners, not spectators.
We do not use invented client counts, savings figures or deployment claims. Our proof is the production software we build and the operating experience behind it.
3 Production AI Products
LLMCap, Selltrix and the AI Recommendation Index are live software products currently offered for sale, with their actual interfaces shown above. AIOptLab, Digitwiner and Underway are explicitly listed as products in development.
Production-minded engineering
Our work connects models to data, controls, interfaces and real workflows so intelligence can be used, measured and improved.
Direct operating experience
The company is led by a practitioner with experience across industrial analytics, decision intelligence, machine learning and production software.
08 / COMPANY
Celikkanat AI Technologies Inc.
A Texas-based technology company that builds its own AI products, applies AI to complex operational problems, and advises companies on decision intelligence and digital commerce. Development is led by Faruk Celikkanat, Founder and Chief AI Officer: a data scientist and technology leader with experience across industrial analytics, operational decision systems, machine learning and production software.
LOCATION
Texas, USA
FOCUS
AI · Data · Automation · SaaS
MODEL
Product Company
PRODUCTS
3 live · 3 in development
LEADERSHIP
Faruk Celikkanat, Founder & Chief AI Officer
09 / INSIGHTS
Two AI agents, one workspace: coordination fails before the code does.
A practical operating pattern for keeping multi-agent software work bounded, reviewable and recoverable.
Read the field note ↗10 / WORK
When we see something in a company's day-to-day work that could be easier, we build a small working sample and send it. Here are three of them. Company names are removed, and the customers, orders and emails in them are invented.
A PDF order form, filled in online.
A herb and spice supplier takes bulk orders on a PDF that customers print and fill in by hand. The sample is the same form in the browser: a missing size is caught before the order goes out, and a list pasted from an email is matched line by line.
Open the sample ↗
A helpline inbox, sorted before anyone opens it.
A wildlife charity answers a shared inbox where a garden question sits next to an injured animal. The sample puts each email into one of three piles with a short reason: ring now, needs a person, or already answered on the website.
Open the sample ↗
Same-day shipping: how many of each on the shelf?
A flat-pack furniture maker wants every product in stock so it ships the day it is ordered. The sample is one page: weekly sales, how much they swing, what is on the shelf and what same-day shipping would take.
Open the sample ↗
11 / CONTACT
Let’s talk.
Tell us about one operational bottleneck, a product question, or a partnership worth exploring.