Growth
2026-10-20 · 7 min read
Sapun Lamichhane
Founder & CEO of Arcetis
Strip away the hype and the daily reality is fairly unglamorous. Marketing teams and small business owners across Nepal use tools like ChatGPT or Claude to draft the first version of a social caption, translate a product description between Nepali and English, summarize a competitor's website, or generate a batch of ad headline variations to test. None of this is transformative on its own - it is closer to a faster intern than a replacement strategist.
Where it genuinely saves time is in the unglamorous repetitive work: writing ten variations of the same offer for A/B testing, cleaning up a messy spreadsheet of leads, or turning a long blog post into five social captions. These are hours saved, not strategy created.
“AI did not remove the need for judgment in marketing - it just moved judgment further up the chain, from writing the post to deciding what the post should say.”
Google's Performance Max and Meta's Advantage+ campaigns now use AI to decide who sees your ad, which creative to show, and how much to bid, often with less manual control than older campaign types gave advertisers. For businesses with a large enough budget and a clear library of creative to feed the system, these can genuinely outperform manually managed campaigns on raw efficiency.
The honest trade-off is control and understanding. These are increasingly black-box systems - you feed in budget and creative and get results back, with limited visibility into why the system made the decisions it made. For a business that wants to understand its own customer, not just its own cost per result, that opacity is a real cost, not just an inconvenience.
A growing number of Nepali businesses - clinics, consultancies, retail shops - now use simple AI chatbots on WhatsApp or Messenger to answer common questions and qualify leads before a human ever gets involved. For straightforward, repeatable questions like pricing ranges or opening hours, this genuinely reduces response time and frees up staff.
Where it backfires is when a business tries to use it to replace the actual human conversation that higher-trust categories depend on. A parent deciding where to send their child for an IELTS consultancy, or a patient asking about a procedure, wants to reach a person fairly quickly. Chatbots that create a wall between the customer and a real human tend to cost more trust than they save time.
This cuts both ways. The same AI tools your team uses internally are increasingly the first stop for your customers too - someone deciding between two clinics in Pokhara might ask an AI tool to compare them before ever visiting either website. This is a shift in consumer behavior, not just a shift in your own workflow, and it means your marketing now has an audience of both humans and AI systems reading the same content.
The biggest danger we see is not businesses failing to adopt AI - it is businesses adopting it carelessly. AI-written content that is not reviewed tends to drift toward a generic, slightly flat voice that is easy to spot and does little to build trust. Worse, these tools will confidently state facts, prices, or claims that are simply wrong if not checked - and in regulated or trust-heavy categories, a wrong claim published under your business's name is a real liability, not a minor error.
The businesses getting real value from AI in 2026 are the ones using it to speed up a human-reviewed process, not the ones letting it run unsupervised.
Start with the internal, lower-risk uses - drafting, translation, data cleanup - where a human reviews the output before it goes anywhere public. Be far more cautious about customer-facing automation in trust-heavy categories, and keep a human easily reachable. If you are weighing how much of your ad management or content to hand to automated AI tools versus keeping in human hands, that is exactly the kind of judgment call Arcetis works through with clients case by case, because the right answer genuinely differs by category and by how much trust your sale depends on.
Not the strategy and judgment part. It has already replaced some of the repetitive execution work - drafting variations, basic translation, simple data tasks - which is pushing agencies to spend more time on strategy, targeting, and review rather than production. Businesses still need someone deciding what to say and to whom.
It depends on your budget and risk tolerance. With a healthy budget and a strong library of creative, automated campaigns like Performance Max or Advantage+ can perform well. With a small budget or very specific targeting needs, the lack of manual control can waste money faster than it saves time.
A general-purpose assistant like ChatGPT or Claude for drafting and translation is the lowest-risk, highest-return starting point. Hold off on customer-facing chatbots until you have a clear sense of which questions are truly repetitive enough to automate safely.