AI sales agent that answers enquiries and qualifies leads
The agent that works while nobody is watching. Most enquiries arrive outside selling hours, and most of those are lost not because the answer was wrong but because nobody answered.
The agent that works while nobody is watching. Most enquiries arrive outside selling hours, and most of those are lost not because the answer was wrong but because nobody answered.
What this agent is for
A sales agent has a different job from a support agent, and configuring it like one is the most common mistake. Support removes contact. Sales creates opportunity — which means its success is not measured by how many conversations it closed without a person, but by how many qualified conversations it handed to one.
The economics are also different. A support conversation that fails costs you a complaint. A sales conversation that fails costs you the sale, silently, and you never learn it happened. That asymmetry should shape how the agent is configured: lean toward capturing and escalating rather than toward containment.
The questions it handles
Is this available?
Specification, availability and lead time — the questions that decide whether someone proceeds or leaves.
How much is it?
Pricing where you publish it, and a route to a person where you do not, rather than a deflection that reads as evasive.
Does it do X?
Capability questions against your own product material, which is where most pre-sale contact actually sits.
How does it compare?
Factual differences against your published material. Not claims about competitors, which is a category of statement no automated system should make.
Can I book a demo?
Capturing the request, the context and the contact detail while the interest is live rather than asking someone to fill in a form later.
I want to talk to sales
Routed immediately, with everything already discussed attached, so the salesperson opens with context rather than with introductions.
A typical conversation
- Someone arrives mid-considerationThey have already read something. The opening question is usually specific, and treating it as the start of a discovery process is how you lose them.
- The agent answers the actual questionDirectly, from your product material, before anything else. Asking for a name and email first is the fastest way to end a conversation.
- It qualifies through conversationCompany, scale, timeframe and use case emerge from the exchange rather than from a form, which is both more pleasant and more accurate.
- It captures what mattersWhere you enable write-back, the enquiry and its context go straight onto the record — while the interest exists, not after someone gets round to it.
- It offers the next stepA demo, a call, a document, a person. One clear step rather than three competing calls to action.
- A person picks it up with contextThe salesperson sees what was asked, what was answered and what was inferred, and continues rather than restarting.
How this agent is configured
| Tone | Direct and informative. Enthusiastic sales language reads as pressure in text, and reads worse in Arabic, where a more measured register carries more credibility. |
|---|---|
| Knowledge | Product and service material, published pricing, case material, and whatever your sales team actually gets asked — which is rarely what the website emphasises. |
| Escalation triggers | Explicit request, high-value signals, procurement or tender language, and any question about bespoke commercial terms. |
| Never states | Claims about competitors, unpublished pricing, delivery commitments it cannot verify, or anything that could be read as a contractual undertaking. |
| Capture | Enquiry, context and contact detail written to your CRM where you enable it, with optional human confirmation before anything is created. |
| Languages | Arabic and English, with pricing and product terminology handled consistently in both. |
Where it hands over to a person
The escalation logic here is inverted compared with support. In support, escalation is a fallback. In sales, escalation is frequently the goal — a qualified conversation reaching a salesperson is a success, not a failure of automation.
That means the triggers should be generous rather than conservative. Procurement language, tender references, questions about bespoke terms, anything suggesting scale — all of these belong with a person quickly, because the cost of a slow response to a serious enquiry vastly exceeds the cost of a salesperson's time on an unqualified one.
The one thing the agent should never do is negotiate. Explaining published pricing is fine. Discussing discounts, terms or anything that could be read as a commitment is a conversation with contractual weight, and it belongs with someone authorised to have it.
What it should never do
Make claims about competitors
Comparative claims are a legal exposure and an automated system is the worst possible place to generate them.
Quote unpublished pricing
If it is not published, it is not the assistant's to state.
Commit to a delivery date
Availability from a connected system is a fact. A promise is a commitment and belongs with a person.
Negotiate
Discounts and terms carry contractual weight. Explaining list pricing does not.
Push
Pressure tactics in text read as desperation, and in Arabic they read worse.
Invent a capability
The most expensive failure in a sales conversation is a confident yes to something the product does not do. It surfaces during implementation, at maximum cost.
Why a separate agent rather than one that does everything
The instinct is usually to build one assistant that handles everything, on the grounds that customers do not care about your org chart. The instinct is wrong, and for a specific reason: tone, knowledge and risk are different per function, and a single agent forces you to average across all three.
Knowledge is the obvious one. The material a support agent should answer from is not the material a sales agent should answer from, and mixing them means each answers from a corpus containing a large amount of irrelevant text. Retrieval quality falls as the corpus widens, so a general assistant is measurably worse at every individual job than a narrow one.
Tone is less obvious and matters more than people expect. The register appropriate for chasing a payment is not the register appropriate for reassuring someone whose service is down. In Arabic this is sharper still, because formality carries more weight and getting it wrong reads as rudeness rather than as informality.
Risk is the one that decides it. Different functions need different limits — what a finance agent may state about an account is different from what an HR agent may state about an employee, and those limits are far easier to set, test and audit per agent than as conditions inside one large one. Separate agents make the question "what is this allowed to say?" answerable.
How this agent should be measured
| Resolution without escalation | Useful, but never on its own. An agent that makes escalation hard scores well here while the relationship deteriorates — and the deterioration surfaces months after the metric improves. |
|---|---|
| Repeat contact within 48 hours | The number that exposes false resolutions. A conversation marked resolved that returns tomorrow was not resolved; it was deferred. |
| Abandonment mid-conversation | Where people gave up. Rises with latency, with menu depth, and with any answer that reads as evasive rather than unhelpful. |
| Escalations requiring repetition | The share of handovers where the customer had to say something twice. This is the single clearest measure of whether handover works. |
| Answer grounding | What proportion of answers trace to a source document. The number a regulator or an internal auditor eventually asks for. |
| Language accuracy | Whether replies came back in the language of the question. The most common complaint about bilingual assistants, and the easiest to instrument. |
| Satisfaction after escalation | Whether people who reached a person were satisfied. If this is high and containment is low, the deployment is working correctly. |
Mistakes worth avoiding when configuring it
Feeding it marketing copy
The questions people ask are answered in policies, procedures and price lists — not on the brochure page. Answer quality tracks the quality of source material more closely than anything else you control.
Writing the answers before the limits
Deciding what the agent must never attempt takes half an hour and prevents the category of incident that cancels projects. Deciding what it should answer takes far longer and matters less.
Optimising for containment
It is the easiest number to improve for the wrong reasons. Make escalation difficult and containment rises while satisfaction falls.
Translating the persona
An English persona rendered literally into Arabic reads as curt; an Arabic one rendered into English reads as ornate. Configure each language rather than translating one.
Launching broad
Narrow deployments that answer a handful of questions extremely well build trust. Broad ones that answer everything adequately lose it at the first confident mistake.
Leaving it unowned
Agents without a named owner drift — material goes stale, tone wanders, and nobody notices until quality has visibly fallen.
What the person on the other side experiences
It is worth thinking about this agent from the customer's side rather than from the configuration screen, because the two views produce different priorities.
From the configuration screen, success looks like coverage: more questions answered, fewer escalations, a rising containment figure. From the customer's side, success looks like something much simpler — they asked, they got a straight answer, and it did not take long. They do not know or care whether the answer came from an assistant or a person, and they will not credit you for automation. They will, however, notice immediately if the answer was wrong, evasive, or in the wrong language.
This is why the honest failure modes matter so much. "I do not have that information, let me get someone who does" costs you a containment point and buys you trust. A confident wrong answer saves the containment point and costs you the relationship, and you will not find out for weeks.
It is also why language handling is disproportionately important here. A customer who writes in Arabic and is answered in English has been told, without anyone saying it, that they are being served by a system built for someone else. In this market that is not a minor usability issue; it is a statement about who the service was designed for.
Typical impact
Setting it up well
Configure it to answer before it qualifies. The instinct is to gather details first; the effect is that people leave. Answer the question that was asked, then let qualification emerge from the conversation. Conversion is consistently better and the data captured is more accurate, because people describe their situation more honestly in conversation than in form fields.
Give it the material your sales team actually gets asked about, not the marketing site. There is usually a substantial gap between what a website emphasises and what buyers ask, and the gap is exactly where enquiries stall.
Wire the capture into wherever your team actually works. An enquiry captured somewhere nobody looks is worse than no capture, because it creates the impression of coverage while opportunities go cold.
Common questions
No, and it should not. Pressure reads as desperation in text and worse in Arabic. It answers, qualifies conversationally, and offers a clear next step.
Yes, where you enable write-back. The enquiry and its context go onto the record while the interest is live, optionally with a person confirming first.
No. Comparative claims are a legal exposure and an automated system is the worst place to generate them. It answers about your own product from your own material.
Published pricing, yes. Unpublished or negotiated pricing is routed to a person, because that conversation carries contractual weight.
It recognises procurement language and escalates quickly. The cost of a slow response to a serious enquiry vastly exceeds the cost of a salesperson looking at an unqualified one.
For complex sales its job is different: answer the qualifying questions, capture the context, and get a person involved fast. Containment is not the objective and should not be the measure.
Configure your own agents
One assistant per function, each with its own voice and limits.