AI analyzer of Yandex and 2GIS cards for lead growth

About the project

An assistant web application has been developed for local business owners and marketers working with Yandex.Maps and 2GIS. The user inserts a link to the organization’s card, the service automatically parses the data (description, contacts, operating hours, photos, reviews, categories, etc.) and sends the structured industrial message to the language model (GigaChat / OpenAI). In response, the application generates a clear report: TL;DR, top 3 priorities for improving the card, detailed steps with justification and expected effect on impressions and leads. This allows you to systematically improve local SEO and increase the conversion of cards into calls and requests.

Business problem

• Automate the collection of data on organization cards from Yandex.Maps and 2GIS.
• Connect a language model that will turn raw data into recommendations for improving local SEO and lead generation.
• Make a clear interface: link → parsing → report with priorities and actions.
• Give marketers and business owners a tool to quickly audit cards.
• Make flexible settings for the AI access mode (direct API, proxy, etc.).

Technical solution

• Field for entering a link to a Yandex.Maps or 2GIS organization card.
• Listing parsing: name, description, contacts, website, social profiles, categories, photos, rating, review count, opening hours, and additional fields.
• Process indication: progress bar “Data collection complete / 100%”.
• Formation and display of “AI logs”: industrial data, request parameters, raw model response.
• Structured report in markdown format: TL;DR, top 3 priorities, detailed recommendations with blocks “What to do / Why is it important / Where to edit / Impact–Ease–Confidence / expected effect.”
• Separate recommendations on work schedule, photographs, descriptions, categories, links, reviews, UTM tags, etc.
• Handling model integration errors (for example, 403 Forbidden when receiving a token) with tips for correcting settings.
• Block “AI Settings”: selection of access mode (direct GigaChat API / via proxy), proxy URL, keys, model parameters.
• Ability to clear logs and run analysis again for different cards.
• The interface is in Russian, tailored for marketers and local business owners.

Technologies

• Frontend: React / TypeScript, responsive web interface
• Backend: Node.js (Express/Nest) for proxying requests to AI-API and data parsing
• Integration with GigaChat API / OpenAI API via custom proxy
• Yandex.Maps and 2GIS page parser (HTTP + HTML parsing / DOM analysis)
• Formation of structured messages and post-processing of markdown responses
• Docker environment for local and server deployment

Results

• The audit time of one organization card has been reduced from 20–30 minutes of manual analysis to 1–2 minutes.
• A unified recommendations format has appeared: clear lists of tasks with priority and evaluation of the effect.
• Marketers can quickly find growth points in cards (work schedule, photos, descriptions, categories, reviews) and increase visibility in search.
• Business owners receive a transparent checklist of actions to increase calls and requests from Yandex.Maps and 2GIS.

Customer review

“The service made it possible to quickly build a clear checklist of improvements for all cards in Yandex.Maps and 2GIS. Now we are systematically finalizing photos, descriptions and work schedules and are seeing an increase in impressions and calls.”