Six systems in production
Every system below is in production as you read this. We describe clients by industry until each agrees in writing to be named. One of the six is our own product.
Hours of drafting, now minutes of review
Regulatory document drafting agent
Hydrogeology engineering firm (EU)
The problem: Regulatory well-project documents took an engineer hours of copy-paste drafting per project, from templates plus site data.
What we built: A production document agent behind an authenticated web app: engineers submit project data, the agent drafts the full regulatory document set in the firm's own format, with background job processing and status polling.
What changed
- 168 regulatory documents generated and 349 boreholes on file in the client database as of August 2026
- Drafting that took an engineer hours now takes minutes of review
- Client came back and commissioned a second system
Claude · Python · LibreOffice · VPS deploy
Stock questions answered in chat
Warehouse stock intelligence agent
Same engineering firm, second contract
The problem: Stock answers lived in an ERP nobody wanted to query. Finding what is in the warehouse meant asking the one person who knew.
What we built: A second production agent for the same client: natural-language answers over live warehouse data, with its own access-controlled portal alongside the document system.
What changed
- Second fixed-price contract from the same client, EUR 3,300
- Both systems run side by side on shared infrastructure
- Deployed with one-command deploy scripts and monitoring
Claude · Python · ERP integration
Ad campaigns managed without a hand on them
Google Ads operations agent
Digital marketing agency (EU)
The problem: Campaign management work was eating senior time: repetitive checks, reporting, and client-by-client housekeeping across accounts.
What we built: An agent that automates the agency's Google Ads operations workflows, with per-client metadata in Supabase and containerized deployment on the agency's VPS.
What changed
- Routine campaign operations run without a human driving them
- Second phase commissioned after the first shipped
- Runs alongside other automation services on shared infra
OpenAI · Supabase · Docker · Google Ads API
Store support in three languages
Multilingual e-commerce support assistant
DTC e-commerce brand (Baltics)
The problem: Support questions in three languages (LT/LV/EE), answered by the founder, at all hours, about products, shipping, and orders.
What we built: A retrieval-grounded chat assistant embedded on the store: product catalog and policy knowledge base, reranked retrieval for accuracy, human handoff for edge cases.
What changed
- Live on the store since 2026, answering in Lithuanian, Latvian, and Estonian
- Answers grounded in the store's own documents, with reranked retrieval
- Founder out of the first-line support loop
GPT-4o · FastAPI · Supabase · RAG + reranking
Our own receptionist, answering real calls
Our own AI receptionist product
In-house product, US market
The problem: The proof most agencies can't show: can you build and RUN a production AI system yourself, at scale, with paying users?
What we built: A full AI receptionist SaaS: 24/7 call answering, website chat, outbound calling, calendar booking, CRM integrations (Zapier, HubSpot, Pipedrive, GoHighLevel, ActiveCampaign), webhook auth and dedup, automated voice evals.
What changed
- Production voice infrastructure we now build client agents on
- Voice eval harness catches regressions before customers do
- It answers our own business line every day
LiveKit · Twilio · OpenAI Realtime · Next.js · Stripe
Book production on a pipeline
Multi-phase AI content production pipeline
International publishing client (via Upwork)
The problem: A multi-stage book production process needing orchestration across drafting, review, and delivery workflows. Too complex for one prompt, too manual as-is.
What we built: A phased n8n + Dify pipeline with separate workflows per production stage, Telegram-based control, and update-safe deployment on the client's VPS.
What changed
- Three phases delivered and extended across follow-on work
- Client operates the pipeline day to day without us
- Survived platform updates thanks to documented update cycle
n8n · Dify · Telegram · VPS
Your system could be next on this page
The first call is free. Paid work starts at the $500 diagnostic, credited in full.