Cases · real systems, not demos

Real work, in operation

Nine systems built and running — in production or in pilot, each stage declared — and each one documented: the problem, the engineering decisions, the stack, and what changed for the business.

Semantic search for industrial catalogs: 200,000+ SKUs found by meaning

Retail · IndustrialIn production

Search that understands context, not just words

Problem

Catalogs with 200,000+ SKUs where exact-word search fails to find what the user actually needs.

Result

The right product is the first result in 65% of searches and in the top three in 90% — even when the user writes another term or another language.

EmbeddingsQdrantpgvectorPostgreSQLDaily syncScalability
Read the full case

Document validation with generative AI for claims and legal teams

Insurtech · Legal (USA)In production

Automated analysis of complex documents, with verifiable citations

Problem

Claims and legal professionals spend hours reviewing long documents to extract critical information and assess risk.

Result

AI does the first read with verifiable citations: 98% agreement with human analysts, measured over two weeks of side-by-side evaluation.

ClaudeGPTOCRPrompt engineeringEncryptionModel failover
Read the full case

Lead generation for insurers with AI and real-time weather data

Insurtech (USA)In production

Business opportunities detected from real weather events

Problem

Identifying properties with potential weather damage by hand arrives late: by the time the sales team reacts, the opportunity has passed.

Result

Leads generated in real time by crossing weather events with geospatial property data. Today it covers three US states.

LLMsWeather APIsPostGISFirebaseMulti-tenantHigh availability
Read the full case

Geolocated marketing: advertising that only reaches people near your business

Retail · CommerceIn pilot

Marketing that reaches only those nearby

Problem

Businesses waste budget on mass advertising that reaches people outside their real service area.

Result

Proximity campaigns with measurable ROI: you only pay for real, location-based reach.

Location APIsNext.jsSocket.ioTypeScriptPrometheusGrafanaObservability
Read the full case

Volumetric dimensioning with computer vision for courier and logistics

Logistics · CourierIn pilot

Instant quotes comparing multiple carriers

Problem

Calculating volumetric weight by hand is slow and error-prone — and every carrier uses a different dimensional factor.

Result

Dimensions by computer vision and instant quotes comparing national and international carriers.

Vision AICarrier APIsNext.jsPythonERP integrationCI/CD
Read the full case

Geolocated alerts for critical operations: warnings that arrive with the phone locked

Industrial · Mining · Oil & GasIn pilot

Knowing when to seek shelter

Problem

Personnel in critical operations cannot tell whether a risk event — lightning storm, incident — actually affects them.

Result

Zone-based alerts that ring with the app closed and the phone locked. Designed for privacy.

IoTWeather APIsModel failoverHigh availability24/7DockerLinux
Read the full case

Real-time sales copilot: AI that assists the salesperson during the video call

B2B Sales · Ambient agentIn pilot

AI that assists the salesperson during the video call, live

Problem

Technical salespeople need specs, prices and account history while they talk — and searching mid-call breaks the conversation.

Result

A copilot that listens to the call and whispers the right fact or question on screen, visible only to the salesperson.

Streaming STTTwo-tier LLMsPrivate vector baseNative appInvisible overlayDual audio
Read the full case

AI-powered ATS: read every CV and rank candidates with judgment, not keywords

HR · RecruitingIn pilot

Read every CV and rank with judgment, not keywords

Problem

Hundreds of CVs per opening: keyword filters discard good candidates who described the same experience in different words.

Result

Every CV read and structured by AI, and an explained ranking: why each candidate stands where they stand.

LLMsStructured extractionEmbeddingsExplainable rankingMulti-format
Read the full case

Intelligent quoting engine: from a pasted free-text list to a formal quotation in seconds

B2B Procurement · DistributionIn production

From a pasted free-text list to a formal quotation

Problem

Quoting a purchase list takes hours: interpreting what the customer wrote, finding each item, comparing prices and building the document.

Result

The customer pastes their list in natural language and receives a formal quotation with market prices in seconds.

LLMsQdrantNext.jsPostgreSQLRedisLLM validationFormal document
Read the full case

Every project includes

UX/UI Design
Full-stack development
LLM integration
Cloud infrastructure
Testing & QA
Technical documentation
Production deploy
Monitoring & alerts
Post-launch support

Have a similar challenge?

Tell us your problem. We'll tell you how to solve it — and how long it takes.