28 FEB, 2026 · media, OOH advertising
AI sales engine and meeting intelligence for a media agency
From referral-based sales to scalable new-client acquisition: ICP-based lead generation, automated outreach and meeting summaries that qualify leads straight into the CRM. All for a running cost of a few tens of dollars a month.
- companies in an ICP-filtered target market database
- 20,000+
- qualified meetings per day (target)
- 3
- monthly cloud cost
- ~30 USD
The bottleneck is sales, not delivery
The agency's professional quality was strong: 40–60% of indicative proposals moved forward. But it won new clients almost only through referrals, while the existing team could have handled a much larger campaign volume. The key to growth was building a systematic lead inflow.
Technical implementation and business impact
Sales engine and domain protection
The system filters target companies by revenue, number of sites and headcount. It distinguishes named target-account lists from general funnel-building communication. Outreach is sent from a separate domain, on dedicated infrastructure, with an open-source mail engine, so the main company domain's reputation stays protected. The AI pre-filters, but leads are loaded only after human approval.
Meeting intelligence and CRM
Instead of subscription note-taking software, we build a lightweight in-house meeting processor. The summary goes into the CRM automatically via API, tagged as a hot lead or with a priority. The CRM is a customised open-source system with client, campaign and partner records, built once and owned by the client.
Measurable business impact and ROI
The conversion chain is fully measurable: outreach → meeting → proposal → order. The target is 3 relevant meetings a day, some of which reach the proposal stage. It runs for ~25–30 dollars a month in a cloud that fits the existing Microsoft ecosystem.
Modular extensibility
We designed the architecture so that the same foundation can later support indicative proposal generation (currently 1.5–2 hours per proposal), an internal knowledge base with built-in assessments (to cut the current 3–6 month onboarding time) and post-campaign reporting. One module launches at a time, and the next follows once it is tuned.