Sales automation with AI: what to automate, what to keep human and how to measure it

    How an SME can build an AI-assisted sales process, from lead qualification to follow-ups, without losing the human touch where it matters.

    HB
    Henrique Baeta
    Co-Founder & Doer
    29 Jul 2026Updated 2 min read
    In this article
    1. What can (and should) be automated
    2. What NOT to automate
    3. A typical stack
    4. How to tell if it is working

    Most sales teams in SMEs spend more time searching for contacts, copying data into the CRM, and writing the same email for the 40th time than selling. It’s not a lack of talent — it’s a lack of a system.

    What can (and should) be automated

    • Lead enrichment (LinkedIn, website, news)
    • Initial qualification via AI agent
    • Automatic summaries of calls and meetings
    • Proposal drafts from templates
    • Scheduled multi-touch follow-ups

    What NOT to automate

    • Negotiation
    • Handling complex objections
    • Relationship building in strategic accounts

    A typical stack

    CRM (HubSpot or Pipedrive) + n8n + OpenAI/Gemini + Apollo + automatic transcription.

    How to tell if it is working

    Before automating, record three numbers: qualified leads per month, time to first response and length of the sales cycle. Measure them again after two or three months. If they don't move, the problem is the process, not the tool.

    It's not magic. It's architecture.

    HB
    Written by
    Henrique Baeta
    Co-Founder & Doer

    Writes about AI consulting for operations and processes: how to map and simplify processes, apply AI where it has the most impact and measure the gains.