Every AI use case, mapped to the business.

32 use cases owners are being sold this year, sorted by where they act in a business of 5 to 100 people. Only 3 are proven in production.

Status
Pick a room to see its use cases
How we rate

Four statuses, one rule: rate the evidence, not the pitch.

Each entry gets a status for how proven it is in businesses of 5 to 100 people, and a separate rating for how strong the evidence behind that status is.

Evidence strength

StrongSeveral independent studies or large independent surveys.
ModerateAt least one independent study or credible survey, plus consistent case evidence.
ThinA handful of independent data points.
Vendor claimsOnly what sellers publish about their own products.

How an entry is made

Each entry is drafted from public sources: academic studies, regulators, independent surveys and reputable reporting. Every figure carries a numbered citation you can open. Anything a vendor says about its own product is labelled as a vendor claim.

Review and updates

Every source in this edition was opened and checked against the text in October 2026. Each entry is then signed off by a named operator and re-checked when the evidence moves. Until that sign-off, the entry is marked as a draft at the top of the page.

The full guide

All use cases at a glance

Field GuideSales & new businessAI outbound sales agents
UnprovenSales & new business

AI outbound sales agents

Also called AI SDR, AI BDR, AI sales development rep, autonomous prospecting agent, AI cold email agent

EvidenceThin
ClassAgent

AI SDRs promise outbound sales on autopilot. We found no independent proof they win business, and real legal and deliverability risks.

ReviewDraft, awaiting operator review
Drafted27 October 2026
Fact-checkedEvery source opened, October 2026
Key takeaway

We found no independent study showing autonomous outbound agents win new business for small firms. The closest field experiment, on structured sales calls, found bots sold well only while undisclosed: telling customers it was a bot cut purchases by more than 79.7%⁠. A large sales-data company said its one-month pilot performed significantly worse than its human SDRs⁠, and in an informal 2024 poll 83% of respondents had got nothing from AI SDRs⁠. In South Africa, cold electronic marketing needs consent⁠.

At a glance

StatusUnproven
What it doesSoftware that finds prospects, writes and sends personalised cold emails or messages, handles replies and books meetings with little human input.
Typical costFrom $1.50 per person contacted on one vendor's pay-as-you-go plan (USD, vendor list price, checked Oct 2026)⁠; most others are quote-based
Data it needsAn ideal-customer profile, a target list with verified business emails and a recorded lawful basis, your proof points, a CRM, and warmed-up sending domains.
Setup effortMedium: separate domains, a two to three week mailbox warm-up⁠, messaging rules and compliance screening come before the first send.
EvidenceThin. One older field experiment on bot sales calls, retail experiments on inbound chat, press investigations and informal polls; no independent study of AI SDR results.

What is an AI outbound sales agent?

An autonomous outbound agent, sold as an AI SDR (sales development representative), does the first stage of new-business work without a person. It builds a list of target companies and contacts, researches each one, writes a personalised email or LinkedIn message, sends a sequence of follow-ups, reads the replies and tries to book a meeting in a salesperson's calendar. Some versions also place phone calls with a synthetic voice.

The pitch to an owner is simple: avoid hiring a junior salesperson, or replace one. The category attracted heavy venture funding; one prominent vendor raised a $24 million Series A and a $50 million Series B⁠. Funding buys marketing, not evidence. This entry rates the category, not any vendor, and asks a narrow question: do these agents produce paying customers for a 5 to 100 person business?

How does it work?

Most products chain the same steps together:

  • A contact database or scraper supplies names, job titles and email addresses that match your target customer.
  • A large language model reads public information about each prospect and drafts a message.
  • A sending engine spreads email across several mailboxes and domains, after a warm-up period of two to three weeks⁠.
  • The model sorts replies, answers simple questions and offers meeting times.
  • Activity syncs back to your CRM.

Volume is the main lever. One media and events business that ran an outbound agent for about six months reported roughly 3,000 AI emails a month, against 75 to 285 from its previous human reps⁠. But Gmail requires bulk senders to authenticate their mail and keep spam complaint rates below 0.3%⁠, so volume and deliverability pull against each other.

What it needs to work

An agent can only send good messages to the right people if you already know who buys from you and why, and can show you are allowed to contact each person.

  1. A written ideal-customer profile drawn from the last 12 to 24 months of won deals in your CRM: industry, company size, buyer job title and typical deal size
  2. A target list with the lawful basis recorded per contact (consent record or existing-customer status for South African recipients and UK sole traders), plus a suppression list of everyone who has opted out
  3. Approved proof points the agent may use: case studies, price ranges and claims you are prepared to stand behind
  4. A separate sending domain with SPF, DKIM and DMARC records, warmed up before launch and monitored for spam complaints⁠
  5. A CRM (HubSpot, Pipedrive, Zoho CRM or Salesforce) where replies, meetings and opt-outs are logged against each contact
  6. A named person who reviews outgoing messages and replies daily for the first three months⁠

Most use cases in this guide succeed or fail here, on the data underneath them, before any AI is involved.

What does the evidence show?

We found no independent study of today's language-model AI SDRs in business-to-business sales. The nearest evidence points in different directions.

A 2019 field experiment with more than 6,200 customers receiving structured outbound sales calls found undisclosed chatbots sold as well as proficient human staff and about four times better than inexperienced ones. Disclosing the bot before the call cut purchases by more than 79.7%⁠. Experiments on a large online retail platform in 2023 and 2024 found an AI pre-sales chatbot lifted sales 16.3%, but that was inbound help for people already shopping⁠. AI-written push messages to platform users had no statistically significant effect on sales, though conversion rose slightly⁠.

The rest is self-reported. One media and events business said outbound AI drove about 10% of ticket revenue for one event, with 15 to 20 hours a week of oversight across its five agents⁠. Against that, an informal poll of a SaaS audience found 83% had got nothing from AI SDRs and 3% had generated real revenue⁠.

AI SDR vs human or founder-led outreach

Compared onAI SDRHuman outreach
VolumeThousands of messages a month⁠Tens to a few hundred personalised messages a month⁠
Message qualityFluent but often generic; can state wrong facts about the prospect⁠Slower, but can draw on real relationships and context
Compliance riskHigh: one mistake repeats across thousands of sends before anyone noticesLower: fewer messages, and errors are easier to catch
Cost modelPer contact or annual contract, plus domains, mailboxes and review timeSalary or the owner's time
Best forLarge markets where you have a clear lawful basis to contact buyersSmall markets where reputation travels and every prospect matters

Status and evidence

StatusUnproven
Evidence strengthThin

Rated unproven because we found no independent study of language-model outbound agents producing customers; the evidence is press investigations, informal polls and self-reported results⁠. Evidence is thin: the one field experiment on bot sales calls predates current tools and shows a large penalty once the bot is disclosed⁠. The large retail experiments concern inbound chat and messages to existing users, not cold outreach⁠.

Where does it go wrong?

  • Consent law. In South Africa, POPIA section 69 bars direct marketing by electronic communication unless the person consented or is an existing customer, and a non-customer may be asked for consent only once. The Information Regulator treats phone calls as electronic communication, so voice agents are covered⁠. Companies count as data subjects, so business lists are not exempt⁠.
  • UK and US rules. In the UK you may email corporate addresses without consent, but sole traders and some partnerships count as individuals⁠. US anti-spam law has no business-to-business exemption, and penalties reach $53,088 per non-compliant email⁠.
  • A burned domain. Breach Gmail's spam threshold⁠ and mail gets filtered; if the agent sends from your main domain, that is where your quotes and invoices come from too.
  • Hidden bots. California bans online bots that hide their identity to encourage a sale; disclosure avoids liability⁠.
  • Invented details and churn. Staff at one vendor said customers complained of hallucinated content; one employee alleged it lost 70% to 80% of customers, while the vendor cited 79% retention⁠.

Numbers it should move

SalesInvoiced sales, month by month.
Quote conversionThe share of quotes that turn into orders.
Customer concentrationHow much of your sales comes from your biggest customers.

Ask the demo business

Northgate Supply is a fictional wholesale distributor of catering and packaging supplies: 28 staff, R46.6m in sales over the last year. Its data is generated for this demo, to 30 September 2026.

Fictional data
How are our sales tracking month by month?
R3.98minvoiced in September 2026

We invoiced R3.98m in September 2026, and R46.6m over the last 12 months. July to September came in 1.2% below April to June. The biggest month was December 2025, at R5.66m, on festive-season trade. Disposables is the largest category, at 22% of sales.

OctNovDecJanFebMarAprMayJunJulAugSepR3.98m
Disposables22%
Cleaning21%
Kitchen equipment20%
Beverage supplies15%
Packaging13%
Paper goods9%
How this was worked out

Invoiced sales excluding VAT, after discounts, per calendar month, and each product category’s share of the 12-month total.

Answered in 5 msScanned 20 493 rowsFrom invoices, invoice lines
Want this on your own numbers, in the AI tools you already use?Book a demo

Where the data usually lives

HubSpotPipedriveSalesforceZoho CRMEmail inboxesExcel and Google Sheets

Where the records this use case needs are usually kept in a small business. It is not a list of recommended tools.

Frequently asked questions

Do AI SDRs actually work for small businesses?

There is no independent evidence yet that they do. The best-known field experiment found bots selling well only when customers did not know they were bots; disclosure cut purchases by more than 79.7%⁠. Self-reported successes exist, but they came with 15 to 20 hours a week of human oversight⁠. Treat any vendor's reply-rate or meeting figure as a claim to test in a small, measured pilot with a clear stop rule.

How much does an AI SDR cost per month?

Public prices are rare. One vendor lists pay-as-you-go pricing of $1.50 per person contacted, with no monthly fee (USD, vendor list price, checked Oct 2026)⁠; most others quote on request. Add the cost of separate sending domains and mailboxes, and above all the staff time needed to review messages and replies. That review time is what made the self-reported successes work⁠.

Do I need a developer or IT person to set up an AI SDR?

Not a developer, but someone comfortable with email infrastructure. Setup means buying separate sending domains, adding authentication records (SPF, DKIM and DMARC), warming mailboxes for two to three weeks⁠, connecting your CRM and writing messaging rules. Gmail expects bulk senders to authenticate mail and keep spam rates under 0.3%⁠. If nobody in the business knows those terms, budget for outside help before launch.

Is AI cold emailing legal under POPIA and UK GDPR?

In South Africa, rarely without consent. POPIA section 69 prohibits direct marketing by email, SMS or phone unless the recipient consented or is an existing customer offered similar products with a chance to opt out⁠. A non-customer may be asked for consent only once⁠, and companies count as data subjects⁠. In the UK, the e-marketing rules (PECR) allow email to corporate addresses but sole traders and some partnerships need consent or an existing-customer exception⁠. See the Information Regulator and the ICO. This is not legal advice.

Do I have to tell prospects they are talking to an AI?

In California it is unlawful to use a bot online to mislead someone about its artificial identity, knowingly deceiving them in order to encourage a sale; a clear disclosure removes that liability⁠. Rules elsewhere vary. Disclosure has a commercial cost: in one field experiment, revealing the bot before a sales call cut purchases by more than 79.7%⁠. Hiding it, though, risks the trust that a small firm's reputation depends on.

More in sales & new business

AI call notesIn pilotsEvidence: ModerateAI lead scoringIn pilotsEvidence: ThinAI proposal draftingIn pilotsEvidence: Moderate

Sources

  1. 1
  2. 2
    Generative AI and Sales Productivity: Field Experiments in Online Retail
    arXiv (Fang, Yuan, Zhang, Donati, Sarvary), 2025Study
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
    Section 1 Definitions - POPIA
    Protection of Personal Information Act 4 of 2013 (text via popia.co.za), 2013Government
  9. 9
    Electronic mail marketing
    Information Commissioner's Office (UK), 2024Government
  10. 10
    CAN-SPAM Act: A Compliance Guide for Business
    Federal Trade Commission (US), 2023Government
  11. 11
    California Business and Professions Code section 17941
    State of California (text via california.public.law), 2019Government
  12. 12

This entry is educational. It is not legal, financial or professional advice, and it does not recommend any vendor. Figures attributed to vendors are their own claims.

Ask your own business. Get the true number in seconds.

Veld connects your accounting, sales and operations data to the AI tools you already use, and keeps it running, so every answer comes from your numbers.

Book a demo