Home / Case studies / AI Sales Automation Platform
AI sales automation platformAI Sales Automation Platform
This AI platform finds leads and client requests for tech specialists across Telegram chats and channels, a shared inbox, websites and job platforms. It turns each post into a structured request and matches every role against the company’s database of engineers and external candidates, with a quote from the CV behind each verdict.
Business challenge
Requests for tech specialists are scattered across Telegram channels and groups, email and job platforms, often cross-posted and phrased differently each time. The client’s sales managers tracked them by hand, then searched a CV folder and a candidate sheet for a fit. Both steps were slow.
We worked alongside the sales team to pin down which channels carry real requests, which details decide a match, and why a wrong match costs more than none. CVs hold personal data, and collected posts are untrusted input, so the platform had to keep both under control.
Our scope included
- Requirements and technical tasks defined with the sales team
- Source connectors and an AI pipeline that turns posts into structured requests
- Evidence-based matching, delivery to Telegram and CI/CD
Product overview
The platform reads Telegram chats and channels, a shared inbox, websites and job platforms. Each source is a connector, so Slack, next on the list, is one new connector rather than a change to the pipeline.
AI separates requests from vendors advertising their own bench and from noise, then turns each request into structured fields: roles, seniority, skills, timezone, rate, duration and English level. A request cross-posted to several channels becomes one record.
Each role is matched against the candidate base. Every verdict (strong, partial or no match) lists fits and gaps, each backed by a quote from the CV. Results arrive as cards in the sales team’s Telegram group, with full details in a private chat.
Values delivered by our team
Forward-deployed engineering
- Requirements defined with the sales managers, from real requests
- “A good match” turned into explicit rules for detection and matching
- Manager feedback fed back into the roadmap
Multi-channel sourcing
- Telegram chats and channels, through a read-only account
- A shared inbox for job alerts and forwarded requests
- Websites and job platforms, with Slack next
Detection and matching
- Requests separated from bench offers and noise
- Cross-posted duplicates merged into one card
- Every verdict backed by a CV quote checked in code
Privacy and platform
- Names and contacts removed before a CV reaches a model
- Read-only access to Drive, Sheets and the inbox
- Modular core, ready for new sources and departments
Other engagements
Let's talk about your project.
Tell us about your goals and we will suggest the right approach and team to get you there.
Most projects can start in under a week.
- 1Tell us what you need
- 2Get a proposal with team, timelines and costs
- 3Start in under a week
We sign an NDA before discussing anything in detail. We reply within 24 hours.