AI can help a sales team by drafting the first version of proposals, follow-ups and pitch materials, so people spend their time on conversations rather than blank pages. It works best as a drafting and organising assistant with a human reviewing everything that reaches a customer. It works badly when it is asked to invent facts, pricing or promises.
Where AI helps and where it does not
| Task | Useful role for AI | What a person must check |
|---|---|---|
| Proposal drafts | Assemble a first draft from your approved descriptions and the client’s brief | Scope, pricing, timelines, every claim |
| Follow-up messages | Draft a polite summary of the call and next steps | Tone, accuracy of what was agreed |
| Meeting notes | Turn rough notes into action lists | Names, dates, commitments |
| Pitch decks | Suggest structure and tighten wording | Facts, brand rules, visuals |
| Research summaries | Summarise a prospect’s public information | Sources and currency |
| Objection handling | Prepare answers from your own documented positions | That answers match real policy |
The pattern is the same each time: AI proposes, a person decides.
Build from a library of approved content
The most reliable way to get accurate drafts is to give the tool your own material. Before you start, collect:
- Service descriptions written and approved by you.
- Standard terms and a list of what is excluded.
- Case examples that are genuinely yours and cleared for use. Never let a tool invent testimonials or results.
- Brand voice notes: words to use, words to avoid.
- Pricing rules, held by a person, not generated.
Then write short prompt templates for each document type that tell the tool to use only the supplied material and to flag anything missing. This idea connects to wider AI automation for business operations, where rules handle the predictable steps and AI handles language.
Keep claims accurate
Sales documents are advertising in the eyes of consumer law. The ACCC says that claims about price, description, value, benefits and quality must be accurate and backed by evidence, that it makes no difference whether a business intends to mislead, and that the overall impression is what counts. That standard applies whether a human or a tool wrote the sentence. General information only: for specific obligations, seek professional advice.
A practical review checklist for any AI-assisted document:
- Is every figure, date and feature traceable to a source you can point to?
- Are there promises of results the business cannot guarantee?
- Does it match the quote, terms and capacity you actually have?
- Are names, company details and job titles correct?
- Does it read like your company, not a generic template?
Google’s guidance on generative AI content makes a similar point in a different context: creators should manually fact check and review all AI-generated content before publishing, because models predict likely wording rather than retrieve facts.
Protect customer information
Proposals and follow-ups contain personal and commercial details. Decide upfront which tools are approved, what may be pasted in and what must not. Review the provider’s terms on data retention and training use, and keep sensitive details out of public tools. If you handle personal information, the Australian Privacy Principles may apply to your business; check the Office of the Australian Information Commissioner’s current guidance or take advice. Customer messages sent in bulk also raise rules on consent and unsubscribing, so confirm requirements before automating follow-up sequences.
A simple rollout
- Pick one document type, such as post-meeting follow-ups.
- Write the template and the approved source material.
- Run it on real work with a reviewer for a few weeks and note every correction.
- Update the template to prevent repeated errors.
- Measure something that matters, like time to send a proposal or response rate, not just volume produced.
- Add the next document type only when the first is stable.
Limits to accept
- Drafts can sound generic and need your voice.
- Tools may state wrong facts confidently.
- Over-automation can make customers feel processed.
- Time saved on writing can disappear into review if prompts are poor.
Pitch material also depends on the strength of your online presence. Prospects check your website before they reply, so see what makes a business website convert, and align sales messages with your wider digital marketing strategy.
Where Radiance Impact fits
Radiance Impact combines creative production, AI, digital technology and marketing, including workflow automation and ready-made automation kits. To talk through a first workflow, see About, the automation checklist or contact us.
Sources
- False or misleading claims, Australian Competition and Consumer Commission
- Google Search’s guidance about using generative AI content, Google Search Central
