{"id":453949,"date":"2026-11-19T15:18:36","date_gmt":"2026-11-19T12:18:36","guid":{"rendered":"https:\/\/menatech.net\/en\/?p=453949"},"modified":"2026-11-19T15:42:32","modified_gmt":"2026-11-19T12:42:32","slug":"ai-is-everywhere-but-is-it-actually-helping-businesses","status":"publish","type":"post","link":"https:\/\/menatech.net\/en\/ai-is-everywhere-but-is-it-actually-helping-businesses\/","title":{"rendered":"AI is everywhere, but is it actually helping businesses?"},"content":{"rendered":"<p style=\"text-align: left;\"><strong>Guest article by Kre\u0161o \u017dmak, Chief Innovation Officer, Infobip<\/strong><\/p>\n<p style=\"text-align: left;\">Almost every business leader can point to an AI pilot or automation project. Yet customers still repeat information across channels, employees still navigate disconnected systems, and too many AI tools answer questions without solving the underlying problem.<\/p>\n<p style=\"text-align: left;\">This matters in the Middle East, where adoption is accelerating. PwC\u2019s 2026 Middle East CEO <a href=\"https:\/\/www.pwc.com\/m1\/en\/ceo-survey\/29th-ceo-survey-middle-east-findings-2026.html\" rel=\"nofollow noopener\" target=\"_blank\">Survey<\/a> found that 39% of Middle East CEOs and 43% of GCC CEOs report extensive AI use across sales, marketing, and customer service, compared with 22% globally. But only 22% of Middle East CEOs and 16% of GCC CEOs believe their most-used AI tools can access all relevant documents and data.<\/p>\n<p style=\"text-align: left;\">The value gap is wider. Deloitte <a href=\"https:\/\/www.deloitte.com\/middle-east\/en\/issues\/generative-ai\/ai-roi-the-paradox-of-rising-investment-and-elusive-returns.html\" rel=\"nofollow noopener\" target=\"_blank\">research<\/a> across Europe and the Middle East found that 85% of organisations had increased AI spending, while only 6% reported payback in under a year.<\/p>\n<p style=\"text-align: left;\">Infobip\u2019s 2026 Customer Experience Maturity <a href=\"https:\/\/www.infobip.com\/cx-maturity\" rel=\"nofollow noopener\" target=\"_blank\">Report<\/a> helps explain why investment does not always translate into impact. While 96% of brands automate customer interactions in some form, only 58% say their channels are fully synchronised, and just 60% have centralised customer-data storage. Only 27% use an orchestration platform, while half say their tools are fully API-ready. Although 53% of brands use agentic AI in customer journeys, deeper adoption is still constrained by user trust (71%), data privacy (64%), and technology-stack integration (41%).<\/p>\n<p style=\"text-align: left;\">The distinction between basic automation and advanced AI is important. Automating a single customer interaction such as sending a fraud alert by SMS is not the same as enabling a two-way conversation in which a customer can resolve an issue immediately. The latter requires connected channels, accessible data, integrated systems, and the ability to orchestrate actions across the customer journey.<\/p>\n<p style=\"text-align: left;\">AI\u2019s capability is not the central issue. The next phase will be defined by whether businesses can connect AI to the data, workflows, and accountability required to solve real problems and deliver measurable outcomes.<\/p>\n<h2 style=\"text-align: left;\">Adoption is not the outcome<\/h2>\n<p style=\"text-align: left;\">Over the past two decades, business communication has evolved from one-way messages to omnichannel engagement and now intelligent, conversational experiences. The technology has changed, but one principle has remained constant: innovation matters only when it removes friction. A chatbot that answers common questions but cannot update an order, process a request, or recognise an existing customer may reduce a few calls, but it does not transform the experience. An AI assistant that summarises a case but cannot connect to the systems where work happens gives employees another screen.<\/p>\n<p style=\"text-align: left;\">Organisations must therefore understand where that friction exists and shift the conversation from what AI can do to what business problem it should solve. The first step is to pinpoint the operational challenges where AI can create meaningful impact: Where are customers abandoning a journey? Which repetitive tasks consume employees\u2019 time? Why are cases transferred between teams? What is driving service costs? The baseline and a business owner accountable for changing it should be clear before AI is introduced. Otherwise, it is impossible to tell whether the technology created value or merely created activity.<\/p>\n<h2 style=\"text-align: left;\">Move from answering to resolving<\/h2>\n<p style=\"text-align: left;\">The next phase of AI will be defined by its ability to resolve problems, not simply generate responses.<\/p>\n<p style=\"text-align: left;\">Generative AI is effective at creating and summarising information. Agentic AI goes further: it can interpret a goal, retrieve relevant context, select the appropriate workflow, and take an authorised action. For example, when a customer asks about a delayed delivery, an AI agent should be able to identify the customer, retrieve the live order status, explain the delay, present available options, update the delivery if requested, and confirm the outcome. If the issue falls outside its permissions, it should transfer the conversation to a human employee with the full context intact.<\/p>\n<p style=\"text-align: left;\">Achieving this requires more than an intelligent model. AI agents need unified customer data, reliable business knowledge, persistent context, connected communication channels, and access to operational systems. Approaches such as agentic retrieval-augmented generation (agentic RAG) can ground responses in approved enterprise knowledge, route questions to the right sources, and check whether the information is sufficient before responding.<\/p>\n<h2>Measure what customers can feel<\/h2>\n<p>Businesses also need to reconsider how they measure success. The number of AI tools deployed or conversations automated says little about business impact.<\/p>\n<p>More meaningful measures include first-contact resolution, time to resolution, cost per resolved case, conversion, abandonment, customer satisfaction, and escalation accuracy. Even containment the share of conversations completed without a human agent must be measured<\/p>\n<p>More meaningful measures include first-contact resolution, time to resolution, cost per resolved case, conversion, abandonment, customer satisfaction, and escalation accuracy. Even containment the share of conversations completed without a human agent must be measured alongside quality. A bot that prevents customers from reaching a person may improve containment while damaging trust.<\/p>\n<p>A useful example comes from one of the Middle East\u2019s online flower-and-gift companies. Using an agentic customer service model built on Infobip\u2019s AgentOS, the company connected AI agents to WhatsApp journeys, company knowledge, real-time CRM data, and human support. At peak demand, it handled 13 times more conversations, reduced its need for peak-period agents by 45%, lowered customer service costs by 15%, and increased customer satisfaction by 12 percentage points year over year. The lesson is broader than any one platform. Value emerged when AI, data, communication, and human expertise worked as a single system. The customer experience and operating model improved together.<\/p>\n<h2>Build for the realities of MENA<\/h2>\n<p>For businesses in MENA and the GCC, language support itself is not evidence of value. An AI system may produce grammatically correct Arabic and still misunderstand dialect, tone, intent, or customers who switch between Arabic and English within the same sentence.<\/p>\n<p>It should be trained and evaluated against real regional conversations and preserve that context across channels and during human handover. Otherwise, localisation becomes another visible capability that fails to improve the customer experience. Businesses should keep authoritative local knowledge current, govern customer data responsibly, and build privacy, consent, data residency, operational resilience, and supplier dependencies into their AI planning. Trust requires clear permissions, auditable actions, confidence thresholds, continuous evaluation, and a reliable path to human support. Human handover is not a failure of AI. At the right moment, it is evidence that the system is working responsibly.<\/p>\n<h2>A more useful AI era<\/h2>\n<p>AI is helping when customers expend less effort, employees can focus on higher-value work, decisions improve, and the business can demonstrate gains in cost, revenue, loyalty, or trust. If leaders cannot point to one of these outcomes, the deployment is not complete.<\/p>\n<p>The companies that lead the next phase will not necessarily be those with the most AI. They will apply it to the right problems, connect it to the right context, and measure it against outcomes that matter. Customers will not remember which model or platform a business used. They will remember whether it understood them and solved their problem.<\/p>\n<p><a href=\"#_ftnref1\" name=\"_ftn1\"><\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Guest article by Kre\u0161o \u017dmak, Chief Innovation Officer, Infobip Almost every business leader can point to an AI pilot or automation project. Yet customers still repeat information across channels, employees still navigate disconnected systems, and too many AI tools answer questions without solving the underlying problem. This matters in the Middle East, where adoption is [&hellip;]<\/p>\n","protected":false},"author":257,"featured_media":453951,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":[],"meta":{"_acf_changed":false,"_breakdance_hide_in_design_set":false,"_breakdance_tags":"","footnotes":""},"categories":[29347],"tags":[29377],"audience-intent":[],"content-types":[],"country":[],"entity":[],"persona":[],"class_list":["post-453949","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-leaders","tag-leaders"],"acf":[],"_links":{"self":[{"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/posts\/453949","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/users\/257"}],"replies":[{"embeddable":true,"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/comments?post=453949"}],"version-history":[{"count":0,"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/posts\/453949\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/media\/453951"}],"wp:attachment":[{"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/media?parent=453949"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/categories?post=453949"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/tags?post=453949"},{"taxonomy":"audience-intent","embeddable":true,"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/audience-intent?post=453949"},{"taxonomy":"content-types","embeddable":true,"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/content-types?post=453949"},{"taxonomy":"country","embeddable":true,"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/country?post=453949"},{"taxonomy":"entity","embeddable":true,"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/entity?post=453949"},{"taxonomy":"format","embeddable":true,"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/format?post=453949"},{"taxonomy":"persona","embeddable":true,"href":"https:\/\/menatech.net\/en\/wp-json\/wp\/v2\/persona?post=453949"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}