AI for Sales & Business Development
AI is absorbing the work around selling: account research, call notes, record updates and first-touch outreach. Sellers keep the conversation, the judgment and the relationship.
How AI is changing this work
- Prospecting becomes an agent's job: research an account, find a reason to write, draft the message.
- Calls are recorded, summarised and scored, so the customer record fills itself in.
- Forecasts lean on what was said in deals, not only on the stage a seller picked.
Your layers: the 3 that matter most
1. Layer 9 · Applications & Interfaces
Why it matters
Sellers will not open another tool. AI that sits in the customer record, the inbox and the call is the AI that gets used.
What to do here
- Assistant in the customer record. Ask for an account brief, the next best step or a follow-up email, written from the record's history. AI features inside products
- Call capture. Meetings are transcribed and summarised; action items and risks land on the deal automatically. Voice & multimodal interfaces
- Coaching from real calls. Review how top sellers handle pricing or objections, using your own recordings. Copilots
Examples as of 2026-11-11
2. Layer 8 · Agents
Why it matters
Outbound work is a loop: find, research, write, follow up. Agents run that loop at a scale no team can, which is both the gain and the danger.
What to do here
- Research agent. Give the agent a question per account, such as a recent funding round, and let it browse and fill the answer in. The agent loop
- Prospecting with approval. The agent drafts outreach from buying signals; a seller approves before anything is sent. The agent loop
- Inbound qualifier. An agent answers first questions on your site and books the meeting for a human seller. Agent permissions & sandboxing
Examples as of 2026-11-11
3. Layer 7 · Orchestration & Tooling
Why it matters
An assistant that does not know your accounts writes generic mail. This layer feeds it the right customer history and connects it to the systems it must update.
What to do here
- Ground it in your records. Connect the customer database, emails and calls so drafts quote real history, not guesses. Retrieval-augmented generation (RAG)
- Write-back as a tool. Let the assistant update fields and log activity through defined actions, never free-form. Tool use & protocols (MCP)
- Account memory. Keep a running summary per account so each new chat starts with what is already known. Memory
Examples as of 2026-11-11
High-leverage use cases
- Account briefs. A one-page brief before every meeting: who they are, what changed, open issues, suggested questions.
- Call notes and follow-ups. Automatic summary, next steps and a draft follow-up email after each call.
- Outbound research at scale. Agents find a specific, true reason to contact each account and draft the first message for approval.
- Clean customer records. Fields are filled from calls and emails, so pipeline reviews start from facts.
- Proposal and tender drafts. First drafts of proposals and questionnaire answers, built from approved past answers.
Start here: the next 30 to 60 days
- Week 1 to 2. Turn on call summaries for one team and compare them with the sellers' own notes.
- Week 3 to 4. Have every seller use an AI account brief before meetings for two weeks, then keep what helped.
- Day 30 to 60. Pilot agent-drafted outreach on one segment with human approval, and track replies against your normal sequence.
- Day 30 to 60. Fix the five record fields the assistant relies on most. Bad records make bad drafts.
Risks and pitfalls
- Invented facts in outreach. A wrong claim about a prospect's company destroys trust in one line. Require a source for every specific claim.
- Scaled spam. Agents can send more than your reputation and your mail domain can bear. Cap volume and keep a human approval step.
- Promises you did not approve. An assistant may offer discounts or terms. Limit what it can state on price, and review customer-facing drafts.
- Recording and privacy rules. Call recording needs consent in many places, and prospect data is personal data. Check both before rollout.
- Telling people it is AI. In the EU, people must be told when they are talking to an AI unless it is obvious. (Regulation (EU) 2024/1689 (the AI Act))
Sources
A product named on this page is an example, not a recommendation.
- Agentforce Sales (Salesforce) (salesforce.com) (2026-11-11)
- Gong (gong.io) (2026-11-11)
- Breeze Assistant (HubSpot) (hubspot.com) (2026-11-11)
- Claygent (Clay) (clay.com) (2026-11-11)
- Prospecting Agent (HubSpot) (hubspot.com) (2026-11-11)
- Salesforce Data 360 (formerly Data Cloud) (salesforce.com) (2026-11-11)
- Zapier MCP (zapier.com) (2026-11-11)
- Regulation (EU) 2024/1689 (the AI Act) (2024)
- NIST AI Risk Management Framework (AI RMF 1.0) (2023-01-26)
Other roles
HR & People · Finance · Marketing · Product & Design · Engineering · Cybersecurity · Legal & Compliance · Operations · Customer Support