AI agents for 

- Scope
- 8 services
- each in its own container
Task: take the routine off the teams — hand market monitoring, reporting, legal review and meeting minutes to agents.
How it works
- 01Competitor monitoringA daily crawl, alerts to Telegram
- 02Task creationTasks into Jira straight from a conversation
- 03Reportsxlsx, pdf, docx from the data
- 04Legal documentsAn independent LLM reviewer
- 05Briefs and minutesFrom meeting transcripts
- 06PresentationsPPTX generated from a brief
- 07Negotiation mapA live assistant during the call
- 08Research on demandSub-agents for the specific task
Particulars and difficulties
Eight independent contours
Each service is isolated in its own container: a failure or an update in one does not touch the others
The size of the task is not known in advance
The agent spins up as many sub-agents as the particular request needs
An agent has to act, not advise
Live integrations with Jira, Telegram and meeting transcripts instead of a text answer in a chat
A result in working formats
Output is xlsx, pdf, docx and pptx — the document goes straight into use without being redone
Results
- 8
- agent services in use
- 4
- document formats out
- 24/7
- monitoring without a person
The market watched every day
Changes at competitors arrive as an alert rather than surfacing at a planning meeting
A second pair of eyes on documents
A legal document passes an independent reviewer before it goes for signature
Stack
LangChain Deep Agents · FastAPI · Docker · Telegram Bot API · Jira