Sites ยท Bots ยท AI automation

Sites, bots, and AI systems for real tasks.

Landing pages, bots, scrapers, catalogs, and AI tools โ€” from an idea to a working result.

Break down the taskView cases
LandingTelegram botScraperAI assistantCatalogDashboard
terminal

process 01 / 05

$ connect sources

ok site: online

ok telegram: online

ok forms: queued

process 02 / 05

$ extract structure

ok schema loaded

ok field map resolved

ok content parser attached

process 03 / 05

$ dedupe records

ok duplicates grouped

ok conflicts tagged

ok empty fields marked

process 04 / 05

$ validate output

ok required fields passed

ok risky records held

ok review queue ready

process 05 / 05

$ publish interfaces

ok site synced

ok bot synced

ok dashboard ready

run progresssystem ready
operator panellive sequence
command 01

Connect sources

Sites, Telegram, forms, catalogs, and boards converge into one intake stream.

output

site: online

output

telegram: online

output

forms: queued

command 02

Extract structure

Text, media, requests, and cards become fields, relationships, and statuses.

output

schema loaded

output

field map resolved

output

content parser attached

command 03

Clean and merge

Duplicates, empty fields, and conflicting records get a clear status.

output

duplicates grouped

output

conflicts tagged

output

empty fields marked

command 04

Validate quality

QA holds risky records before publishing and shows where manual review is needed.

output

required fields passed

output

risky records held

output

review queue ready

command 05

Publish outputs

The clean stream lands in a site, Telegram bot, API, catalog, or operator panel.

output

site synced

output

bot synced

output

dashboard ready

terminal

sergei-system:~$ run dirty-data-loop

[01] connect --sources site telegram forms catalog

ok site: online

ok telegram: online

ok forms: queued

[02] extract --llm --schema task_loop --media

ok schema loaded

ok field map resolved

ok content parser attached

[03] dedupe --strict --merge-conflicts

ok duplicates grouped

ok conflicts tagged

ok empty fields marked

[04] validate --qa-gates --before-publish

ok required fields passed

ok risky records held

ok review queue ready

[05] publish --site --bot --api --dashboard

ok site synced

ok bot synced

ok dashboard ready

done: loop synchronized

run progresssystem ready
Break down the taskView the proof
operator panel
command 01

Connect sources

Sites, Telegram, forms, catalogs, and boards converge into one intake stream.

output

site: online

output

telegram: online

output

forms: queued

command 02

Extract structure

Text, media, requests, and cards become fields, relationships, and statuses.

output

schema loaded

output

field map resolved

output

content parser attached

command 03

Clean and merge

Duplicates, empty fields, and conflicting records get a clear status.

output

duplicates grouped

output

conflicts tagged

output

empty fields marked

command 04

Validate quality

QA holds risky records before publishing and shows where manual review is needed.

output

required fields passed

output

risky records held

output

review queue ready

command 05

Publish outputs

The clean stream lands in a site, Telegram bot, API, catalog, or operator panel.

output

site synced

output

bot synced

output

dashboard ready

proof / mechanics

Proof, not promises

Family Tree demonstrates cards, relationships, and media. Facebook Scraper demonstrates data collection and structure. BanBao demonstrates a working marketplace, catalog, and AI assistant.

BanBao home pageBanBao local listings catalog
Flagship caseIn development

BanBao: a local marketplace for Nha Trang

A working marketplace and assistant for local offers: food, real estate, services, and listings sourced from Facebook, Zalo, and other channels. It brings catalog, translation, search, and seller contact into one product.

Open BanBaoOpen beta: the product works and continues to evolve
MarketplaceNha TrangCatalogAI assistant

Other working projects

Tree
Illustrated landing screen for the Lesnov family tree
PeopleRelationshipsStoriesMedia
DemoOpen

Family Tree Demo

People cards, relationships, media, and a public product interface.

Product UIFamily treeSupabaseVercel
Groups
Facebook Group Posts and All Photos Scraper page on Apify
texttimestampauthorengagementpostUrlphotos[]
+NJSON
Apify ActorOpen

Facebook Posts + All Photos Scrapers

Collection, structuring, and export of posts from Facebook groups and public profiles.

ApifyStructured JSONMonitoringHidden photo grids
Demo
TG-AI interactive demo with AI personas inside Telegram
RolesContextPausesTelegram
Interactive demoOpen

TG-AI

AI personas and controlled scenarios for Telegram groups, comments, and direct messages.

TelegramAI personasScenariosProduct UI
VPS
Linkit Cloud landing page offering European VPS hosting
VPSPricingComparisonSetup
Live landingOpen

Linkit Cloud

A European VPS landing page with plans and optional managed service setup.

LandingVPSPricingConversion UI
sergei-system:~$ match --task --working-mvp

common briefs

Does this look like your task?

The same product mechanics repeat across matching services, booking, bots, dashboards, and internal systems. Build the core user journey first, then add integrations and growth.

Roles

Profiles

Search

Requests

Notifications

Moderation

01

Matching service / marketplace

Roles, profiles, filters, requests, reviews, and moderation.

02

Telegram bot + web panel

Requests, statuses, notifications, and a web control panel.

03

Booking, scheduling, and orders

Services, schedules, slots, rescheduling, and reminders.

04

Scraper and monitoring

Sources, filters, deduplication, and notifications.

05

AI assistant over your data

Answers over documents, catalogs, or an internal knowledge base.

06

Internal dashboard / CRM

Users, statuses, tables, control, and export.

runtime manifest

Stack

The stack follows the task: the first working release, its data, integrations, and the path to grow it.

OpenAICodexLLM pipelinesstructured extraction
PythonPlaywrightApifyBeautifulSoupPostgres
TypeScriptNext.jsTailwindshadcn primitivesSupabase
Telegram APIDockerGitHub ActionsVercel

Contact

Have a similar task? I will turn it into a clear first release.

Describe the users, core journey, data sources, and what absolutely must work in the MVP.

Break it down in Telegram

Describe the task, users, data, and must-have features for the first version.