Documents & language
AI agents · RAG · chatbots · voice assistants
Assistants and agents that answer from your own documents, tickets and logs, with citations and an accuracy number you can rerun.
- Industries
- Insurance · Finance · Telecom support
Applied AI studio · Seoul
We turn messy documents, images and device data into AI systems your team actually uses, and put them to work inside your website, CRM and apps. A working prototype in weeks, measured against a number you can check.
What we build
Every project is the same problem underneath: finding the signal in messy data. We’ve solved it in documents, images, and sensor signals, so methods proven in one industry transfer directly to yours.
AI agents · RAG · chatbots · voice assistants
Assistants and agents that answer from your own documents, tickets and logs, with citations and an accuracy number you can rerun.
Inspection · safety compliance · authentication · medical imaging
Detection and segmentation models that find defects, violations and anomalies in the images and video you already capture.
Telemetry · fault & anomaly detection · on-device AI
Models that read logs, waveforms and sensor streams the way your senior engineers do, on a server or on the device itself.
Put it to work
We find the signal in your data, then put it to work inside the tools your team already uses, with guardrails and human approval where it matters.
Answers visitors, qualifies leads and routes them with context.
Enriches, scores, logs and drafts follow-ups for approval.
Copilots, search, extraction and reports inside your product.
Across all three: LLM integration & advisory. Model selection, prompting vs. fine-tuning, private deployment and evaluation: an honest engineering view before you commit budget.
LLM advisoryTrack record
Our team has shipped production systems and published peer-reviewed research across documents, images and signals. Every project below has a written case study, labeled honestly as client work or R&D.
Telecom · home networking
Vague “my internet is slow” tickets become a ranked root cause, the log lines that prove it, and the next action.
Result: 7 analysis passes · 3 vector memories
View case studyInsurance · South Korea
Customers get answers from the insurer’s own policy documents, by chat or by phone, instead of generic chatbot guesses.
Result: Korean end to end · chat + phone
View case studyInfrastructure inspection · South Korea
One model outlines five kinds of concrete damage, so inspectors measure defects instead of eyeballing photos.
Result: 5 damage classes · 5 models benchmarked
View case studyIndustrial safety
Site cameras check seven types of protective equipment automatically, so safety teams don’t have to watch every feed.
Result: 7 PPE classes · 2.3k training images
View case studySmart city · traffic & parking
Reads plates from the low-quality street CCTV operators already have, with no camera upgrade needed.
Result: >90% recognition accuracy
View case studyRetail · brand protection
Automated first-line screening of genuine vs. counterfeit logos for brand-protection teams and resale platforms.
Result: 94.8% AUC-ROC · 94.0% F1 (brand A)
View case studyIndustries
Client projects, research and peer-reviewed work across nine industries. Pick yours to see the relevant projects, or read an industry guide.
Proven here, applied there
You don’t need a vendor who has built your exact system before. You need one who has solved the same underlying problem and can show the evidence.
Who you’ll work with
The people you first talk to are the people who scope the work, write the code and answer your questions after launch. No account managers, and no hand-offs to juniors.
Cofounder & Principal AI Engineer
PhD, Sungkyunkwan University
SaeroSense joins 새로 (saero), Korean for “new,” with sense. The idea behind it: AI that gives you a new sense of the data you already have, whether it lives in documents, images or sensor signals.
Dr. Saad has delivered industrial and academic AI projects, from carrier-facing telecom systems and visual inspection to medical imaging research, and his peer-reviewed work is published in IEEE Transactions on Power Delivery.
No juniors learning on your budget and no outsourcing. Our engineers, led by Dr. Muhammad Saad, work with you from the first conversation to handover.
Every engagement is scoped against a metric and ships with the harness that proves it, so “it seems better” becomes a number.
No reseller agreements. Model and platform choices follow your requirements and measured results, not a commission.
Code, models, evaluations and runbooks are yours. And if you’d rather not run it yourself, we operate it for you.
How an engagement runs
Most AI projects fail between the demo and the deployment. Our whole method is built around that gap: fixed prices per phase, real data from week one, and an evaluation you can rerun after we leave, or that we rerun for you every month.
Less guesswork. More shipped systems.
We study your data and workflow, define the metric that matters, and give you a plan with a fixed price.
A working system on your real data, not slides, with an evaluation you can rerun yourself.
Hardening, deployment to your cloud or on-premise hardware, monitoring, and training for your team.
Documentation, runbooks and training. You own the code and the models.
We monitor, re-evaluate and improve the system every month, so it stays accurate as your data changes.
Engagement models
Three ways to work with us. Most clients start with Scope, move to Build once the numbers say it’s worth it, and add Operate if they’d rather not run the system themselves.
Fixed price · 1 to 2 weeks
We study your data and workflow, agree on the metric that defines success, and give you a fixed quote for the build.
Fixed price · prototype in 2 to 6 weeks, then production
A working prototype on your real data, then hardening, deployment and handover to your team.
Monthly · cancel with 30 days’ notice
We keep your system accurate, current and cheap to run, so your team doesn’t have to.
Working with us
We’re available throughout US business hours, Eastern to Pacific. And because our team is in Seoul, work also moves forward while you sleep.
Every project starts with a short scoping phase at a fixed price. After that you get a fixed quote for the build, and an optional monthly Operate plan if you want us to run the system. We don’t bill by the hour, so there are no surprise invoices. See how we work.
Yes. With the Operate plan we monitor the system, rerun the evaluation every month and send you an accuracy report, fix drift, upgrade to newer or cheaper models, and give you priority support. You can cancel with 30 days’ notice, and you keep owning everything.
We’re available throughout US business hours, Eastern to Pacific, and we reply within 1 business day. Because our team is in Seoul, work also moves forward while you sleep. Everything is in English: proposals, NDAs, documentation and code.
We sign an NDA before you share anything. Systems can run entirely in your cloud or on your own hardware with open-weight models, so sensitive data never goes to an outside API. You own all code, models and evaluations, and we delete your data when the project ends.
Send two or three sentences about your data and the outcome you want, through our contact page or to contact@saerosense.com. We reply within 1 business day with an honest read on feasibility, and if we’re not the right fit, we’ll say so.
Start a project
Two or three sentences about the data you have and the outcome you want is enough. We reply within 1 business day with an honest read on feasibility. No slide deck needed.