When businesses should — and should not — use AI agents

Use an agent when these are true

• The same questions arrive every day on WhatsApp, email, or Instagram
• Leads after hours go cold because nobody is at the keyboard
• Quotes sit unchased because follow-up depends on memory
• Someone can approve tone, prices, and handoff rules in writing
• Every serious enquiry should become a CRM record with a stage

Do not start with an agent when

• Nobody can say which packages may be quoted on WhatsApp
• The public website is the first leak — unclear offer or missing trust pages
• You want a homepage widget while real demand lives in DMs you refuse to connect
• The work is mostly one-off judgment (bespoke design exceptions, admissions edge cases) with no policy

Industry-shaped examples (without invented KPIs)

Real estate teams drowning in ad leads need qualification and CRM writes before a closer joins — see /en/industries/real-estate. Kitchen studios with three messaging inboxes need one pipeline; Ready Kitchen is Veyrox’s in-progress example on /en/work/ready-kitchen and /en/industries/kitchen-remodeling. Restaurants often need brand sites first, then enquiry routing — F&B demos live under /en/work and /en/industries/restaurants.

How the free audit decides

Bring the inbox or URL. We map whether an agent, a website, CRM hygiene, or waiting is the honest first step. For the broader business overview of agents, see AI agents for business; when you are ready to build, see AI agents and automation services. ROI is modelled before paid build. Book at /en/contact or read /en/blog/free-ai-audit-30-minutes.