Stack · AI & data
LLM & agents
Agentic AI in production: agents, chatting with documents, RAG, structured extraction and classification.
3 places2 challenges
My experience
Filetable's pipeline (classification, structured extraction, traceability of every value), Branchlate's translation suggestions, and Swelv's recommendations, where the model searches the catalogue itself.
Where I used it
The challenges
- Queues and priorities: orchestrating thousands of documents with TemporalA document goes through a dozen steps of OCR, AI and SQL. The real challenge is not chaining them, it is deciding who goes first.7 min
Filetable - A RAG without vectors: the model finds its own booksEvery month, five books picked for each reader. The model invents no title: it queries the catalogue with tools, and code checks everything.6 min
Swelv
In the same family
Claude CodeClaude CodeMy everyday tool to write, test and review code, framed by written rules and never shipped without review.Used atFiletableSwelvSee the details →OCROCRReading scans and photos, automatic splitting of multi-document PDFs.Used atFiletableSee the details →PostgreSQLPostgreSQLFlexible relational models, migrations without downtime, traceable and auditable data.Used atFiletable
BranchlateSwelv
OneDoc
ID QuantiqueSee the details →
FiletableSwelvSee the details →OCROCRReading scans and photos, automatic splitting of multi-document PDFs.Used atFiletableSee the details →PostgreSQLPostgreSQLFlexible relational models, migrations without downtime, traceable and auditable data.Used atFiletableBranchlateSwelvOneDocID QuantiqueSee the details →