OLX Agentic Selling and DealFinder: AI With Humans in Charge

Direct answer: OLX Group has launched two new AI marketplace tools, Agentic Selling for sellers and DealFinder for buyers, with a clear design principle running through both: AI should handle the complexity of a listing or a search, while people keep control of price, disclosure, and the final decision. For UK small businesses that sell through online marketplaces, the launch matters less as a product you can sign up for today and more as a signal of where marketplace selling is heading — towards guided, conversational, agent-assisted journeys in which trust, transparent pricing, and honest disclosure become the competitive edge.

What happened

OLX Group unveiled Agentic Selling and DealFinder at CLAIM Online 2026, its invite-only summit on the future of marketplaces, on 27 October 2026. The formal press release followed on 28 October 2026, setting out the two tools alongside new survey research into how customers actually feel about AI in buying, selling, and job hunting. The launch was reported in the trade and technology press the same week, including coverage summarising the announcement for readers following agent-led commerce.

To understand why this launch carries weight, it helps to understand the scale of the company behind it. OLX Group describes itself as a global online classifieds leader, with nearly 60 million daily listings across 11 brands in seven markets. It is the classifieds business of Prosus and serves Europe and South Africa, with deep roots in motors, real estate, jobs, and general classifieds. That footprint matters because marketplace tools are only as good as the data and dealer relationships behind them. OLX points to more than 20 years of dealer and realtor relationships and verified marketplace data, and to more than eight years of building AI into its marketplaces. The result, in its telling, is not a sudden leap into fashionable AI features, but a gradual accumulation of capability that has now reached the point where the company feels able to hand parts of the buying and selling journey to agentic systems.

The numbers OLX cites for its AI estate give a sense of that accumulation. The group reports 86 live Machine Learning use cases, 28 live Generative AI use cases, and 14 agentic solutions already in the market. Those figures are worth pausing on, because they cut against a common pattern in technology announcements: a single flagship feature presented as if it appeared from nowhere. Here, the two new tools sit on top of a much wider programme of AI work, much of it invisible to the end user — ranking, fraud detection, pricing signals, content generation, and matching — that has been tested in live marketplaces over time. For a small business owner, that distinction is useful. A tool that has been refined through dozens of live use cases is more likely to behave predictably than a tool launched in isolation, and predictability is exactly what you need when the tool is touching your prices and your reputation.

The announcement was framed, unusually, around trust rather than automation for its own sake. OLX paired the product news with two October 2026 surveys, and the survey findings read less like marketing support and more like a corrective to the industry’s enthusiasm. In the jobs survey, carried out by SW Research among 1,510 people aged 18 to 29 in cities of 200,000 or more — around 300 each in France, Germany, Poland, Romania, and Ukraine — between 92% and 96% of respondents said they use AI tools personally, yet fewer than 41% felt ready for the labour-market changes AI is bringing. Concern about AI careers stood at 31% for both men and women, with Poland the most cautious at 37% reporting high or very high concern. In other words, familiarity is not the same as confidence. People are using AI every day and still feeling unprepared for what it means for their working lives.

The real estate survey told a similar story from the other side of a transaction. Among 2,819 respondents — completed and incomplete responses from home seekers on OLX platforms in Poland, Portugal, and Romania — the most requested features were alerts and side-by-side comparison, practical tools that help a person decide, rather than tools that decide for them. Only 5% said they would trust an AI valuation over a human agent’s estimate, while 36% said they trust the realtor’s estimate. And 78% said they wanted AI-altered listings to be disclosed. OLX is careful to note that this survey is not representative of all home seekers, but the direction of travel is hard to miss: customers want help, they want comparison, and they want to know when AI has touched what they are looking at.

That is the context in which Agentic Selling and DealFinder make sense. OLX’s senior leaders framed the launch in those terms. Chief Data Officer Andreas Merentitis argued, in essence, that AI should take on complexity while people make the decisions that matter, and that trust is what makes the technology useful in the first place. Chief Executive Christian Gisy made a related point: counting how many people have tried AI tells you about adoption, not about trust in the outcome. People want to see why a tool recommends something, and they want to keep the final call. Paraphrased like that, the two messages sound almost obvious. In practice, they are a design commitment, because it is considerably easier to build an AI that simply picks a price or fires off a message than it is to build one that explains its reasoning, shows its workings, and then steps back.

Key facts at a glance

Timeline

• February 2026: CLAIM Lisbon, the in-person edition of OLX’s future-of-marketplaces summit, takes place.
• 27 October 2026: OLX unveils Agentic Selling and DealFinder at CLAIM Online 2026.
• 28 October 2026: The formal press release announcing the two tools is published.

The summit itself deserves a brief note, because it explains the tone of the launch. CLAIM is an invite-only event, and the 2026 online edition followed CLAIM Lisbon in February. Speakers listed for the event included Sam Burman, Aleksandra Przegalinska, Gertrud Kolb, Martin Alcala Rubi, and Joana de Quintanilha — a mix that reflects the summit’s positioning at the intersection of technology, research, and marketplace practice, rather than as a pure product showcase. That positioning fits the announcement: this was presented as a statement about how marketplaces should use AI, not merely as two new buttons in an app.

It is also worth being precise about what this launch is not. OLX does not operate a UK consumer marketplace in the way it operates in Poland, Portugal, or South Africa, so a UK estate agent or car dealer cannot simply switch on Agentic Selling or DealFinder tomorrow. The value of the story for a UK reader lies elsewhere: in the pattern it reveals, in the survey evidence about customer attitudes, and in the practical standards — transparent pricing guidance, seller control, disclosure of AI-altered content, conversational buyer assistance — that UK buyers and sellers are increasingly likely to expect wherever they trade. Marketplaces tend to converge. When one large operator normalises a feature, customers carry that expectation to the next platform they use, including the ones you sell on.

How Agentic Selling and DealFinder actually work

Agentic Selling is the seller-side tool, and its job is to walk a private seller through the whole journey from valuation to a live listing. In OLX’s description, it guides sellers from valuation to posting, provides transparent, data-backed pricing guidance, and produces listing drafts. The crucial detail — and the one that separates it from a fully automated listing machine — is that final pricing and disclosure decisions stay with the seller. The AI proposes, explains, and drafts; the human approves, adjusts, and publishes.

Think through what that means in practice, step by step, because the workflow is where the value lives. A seller who is unsure what their item, vehicle, or property is worth no longer starts with a blank price field and a guess. Instead, the system can draw on verified marketplace data — what comparable items have listed for, how quickly they moved, what condition and specification details tend to matter — and turn that into guidance the seller can actually inspect. Transparent, data-backed guidance should mean more than a single number on a screen. At its best, it means a range, the reasons behind it, and the factors that would push the price up or down. A seller who can see that reasoning is in a position to disagree intelligently, to adjust for something the data cannot see — a recent refurbishment, an unusual feature, a local quirk — and to set a price they can defend when a buyer asks why.

The next step, the listing draft, is equally important and equally easy to underrate. Writing a good listing is skilled work. It requires a clear title, an honest description, the right specification details, and photographs that show condition rather than hiding it. Many private sellers stall at exactly this point, or they publish thin listings that attract low-quality enquiries and wasted viewings. An AI that drafts a title and description from the seller’s inputs, and from what it knows about the category, removes that friction. OLX’s wider results suggest this is not just theory: the group reports that AI-generated titles and descriptions lift posting completion by 11 percentage points. That is a meaningful figure, because completion is the bottleneck. A listing that never gets published sells nothing, however good the price might have been.

Then comes the decision point, and this is where OLX’s design philosophy is tested. The seller reviews the draft, checks the price guidance, decides what to disclose, and publishes — or does not. Keeping that decision with the seller is not a cosmetic choice. It protects the seller from an AI confidently mispricing something it does not fully understand, and it protects the buyer from a listing that presents an AI’s guess as a fact. It also preserves accountability. If a price is wrong, or a disclosure is missing, there is a person responsible, not a system to blame. For anyone who has ever dealt with an automated system that made a confident mistake on their behalf, the appeal of that arrangement should be obvious.

Agentic Selling is currently live for 20% of private sellers on Otodom in Poland. That limited rollout is itself informative. It suggests a phased approach — test with a slice of sellers, learn from real behaviour, refine the guidance and the drafts — rather than a big-bang launch across every market at once. Phased rollouts are often read as caution, but in marketplace tools they are usually a sign of seriousness. Pricing guidance that is slightly wrong at scale damages trust quickly, and trust, as OLX’s own research shows, is the scarce resource.

DealFinder works from the opposite side of the transaction. It is a buyer-side assistant that lives on WhatsApp, and it is being scaled after a beta in Portugal with 2,000 users. The buyer describes what they are looking for, and DealFinder finds items, sets price and item alerts, and accepts searches by text, image, or voice. In the beta, it achieved an 18% conversion rate to contact — that is, nearly one in five beta users went on to contact a seller. For a buyer assistant, that is a striking early signal, because the hardest part of any marketplace search is not finding listings, but finding the right listings and then taking the step of making contact.

Consider the buyer journey it changes. Traditional marketplace search is a filter exercise: choose a category, set a price range, tick boxes for specification, scroll, refine, repeat. It rewards people who already know the terminology and the trade-offs of the category, and it punishes everyone else with either too many results or none at all. A conversational assistant on WhatsApp inverts that. The buyer can write, in ordinary language, what they need, send a photograph of something similar to what they want, or simply say it out loud. The assistant translates that intent into a search, watches for new matches, and alerts the buyer when something fitting appears or when a price moves. For a busy buyer — a parent looking for a car, a renter scanning for a flat, a small business sourcing equipment — that shift from active searching to monitored searching is a genuine change in effort, not just a new interface.

The choice of WhatsApp as the channel is also worth noting. Buyers already live in messaging apps, and a marketplace assistant that meets them there removes the need to open a separate app, remember to check it, and repeat a search. This is part of a wider move towards conversational commerce that UK businesses should watch closely, because messaging is where many customer relationships now happen. The principle is explored in more depth in guidance on AI agents live in text messages, WhatsApp and SMS, which looks at how agent-style help inside messaging changes what customers expect from a business that answers them there. DealFinder is a marketplace-scale example of the same idea: help that arrives in the thread, in the buyer’s own words, rather than behind a separate login.

The two tools are also connected to a broader set of live AI results that OLX cites, and those results help explain why the company believes the moment is right for agentic features. AI-powered ad posting cuts time-to-post by 50% in real estate, according to OLX. In Motors, more than 70% of dealers accept AI-made videos. Agentic search is live across Motors and Real Estate. AutoIQ, a dealer agent, is live with all dealers in Poland and Portugal. In Jobs, an AI Interviewer runs more than 1,000 practice interviews a month, and a CV Optimiser has drawn positive early-user reviews, with 88% of users finding it useful and 90% saying they love the AI Interviewer. None of these figures, on their own, proves that Agentic Selling or DealFinder will succeed. Together, however, they describe a pattern: AI assistance is being accepted, in specific tasks, by dealers, sellers, and jobseekers who have something concrete to gain from it — faster posting, better listings, more practice, clearer applications. Acceptance in a narrow task is often the precursor to trust in a broader one.

That pattern is the real story of this launch. Agent commerce is often discussed as if it means handing a purchase to a bot and stepping away. OLX’s version is more restrained and, for that reason, more plausible as a near-term future: agents that prepare, propose, monitor, and draft, while people approve, adjust, and decide. The seller still prices. The buyer still contacts. The agent does the searching, the drafting, and the watching in between. If that model spreads — and the early signals suggest it will, in one form or another — then the skills that matter for sellers will shift. Writing a perfect listing from scratch will matter less. Checking an AI draft honestly, pricing transparently, disclosing fully, and responding quickly to a well-matched buyer will matter more.

What it means for UK small businesses

The most important shift here is not a single feature. It is a change in who does the work of buying and selling, and in what buyers will soon take for granted. Until recently, a marketplace listing was a static page: photographs, a description, a price, and a contact button. The buyer did the searching, the comparing, and the chasing. Agentic tools turn that page into a guided process. On the seller side, the system helps you price and present. On the buyer side, it keeps watching after the buyer has walked away, and it taps them on the shoulder when the right thing appears. For a UK small business, that changes the job in four practical ways: your listings need to be machine-readable as well as human-readable, your pricing needs to be defensible, your response speed matters more, and your honesty about AI use becomes part of your brand.

Start with the listing itself. If buyer assistants are going to search by text, image, and voice, then the details that help an assistant match your listing to a real person matter as much as the headline. That means complete, accurate specification fields, photographs that show the item or property from the angles a buyer would actually check, and descriptions that state condition plainly — including flaws. It also means avoiding the temptation to stuff a title with keywords that do not describe the item. An assistant that matches on meaning, rather than on keyword overlap, will be less fooled by that tactic, and a buyer who arrives via a conversational assistant will arrive with a clearer expectation of what they are about to see. A listing that disappoints that expectation costs you not just a sale, but a review, a message thread, and time you cannot get back.

Pricing is the second shift, and it is the one OLX is most explicit about. Transparent, data-backed pricing guidance only works if the seller understands it and can explain it. UK sellers can adopt the same discipline today, with or without OLX’s tool, and this is general business advice rather than a claim about any platform. Before you list, gather your own evidence: recent comparable sales, how long similar items stayed listed, what condition differences justified a higher or lower price, and what your own costs and margins require. Write that reasoning down, even if only for yourself. When a buyer asks why the price is what it is — and buyers assisted by AI will ask more precisely, because they will arrive armed with comparisons — you will have an answer that is specific, calm, and credible. If you use an AI pricing suggestion on any platform, treat it as a starting point to check, not as a verdict to accept. Check it against your comparables, adjust for anything the system cannot see, and be ready to explain the final figure in one or two sentences. That habit protects your margin and your credibility at the same time.

Disclosure is the third shift, and the survey evidence makes it unavoidable. If 78% of home seekers in OLX’s real estate survey want AI-altered listings disclosed, then disclosure is not a niche concern for the cautious few. It is a mainstream expectation. The practical rule for a UK seller is simple: if AI materially changed a photograph, a description, a video, or a valuation in your listing, say so, briefly and plainly. You do not need a legalistic paragraph. A short note — that images were enhanced, that a description was drafted with AI assistance and checked by you, that a video was generated from photographs — is enough to preserve trust, and it costs you nothing in presentation. What damages trust is the opposite: a buyer discovering the alteration themselves, after they have travelled to a viewing or committed to a purchase, and wondering what else was altered without being mentioned. In property in particular, where trust is the product as much as the house, that distinction matters enormously. Estate agents thinking through how AI fits a traditional, relationship-led sale will find useful context in a UK-focused guide to AI for estate agents, which sets out where automation helps an agency and where the human relationship still does the selling.

Response speed is the fourth shift, and it follows directly from DealFinder’s beta result. An 18% conversion rate to contact means a significant share of assisted buyers are not browsing idly; they are reaching out. When a buyer assistant has already filtered, compared, and alerted, the contact that follows is warmer than a cold enquiry from a search-results page. That warmth decays quickly if nobody replies. A small business that treats marketplace messages as something to check once a day will lose those buyers to a competitor who replies within the hour, with a specific answer rather than a generic acknowledgement. This does not require being online around the clock. It requires a simple routine: check messages at set times during the day, answer the actual question asked, confirm availability and viewing or collection options, and keep the conversation in the same channel the buyer used unless they ask to move it. If you use messaging for customer contact more broadly, the same discipline applies — a fast, specific, human reply beats a fast automated reply that answers nothing.

A playbook for estate agents

For estate agents, the OLX story is a preview of a listing workflow in which valuation guidance, drafting, photography, and video are increasingly assisted, but the agent’s judgement remains the point of differentiation. The survey finding that only 5% would trust an AI valuation over a human agent’s estimate, against 36% who trust the realtor’s estimate, should be reassuring rather than threatening. It suggests that buyers still want a professional who knows the street, the building, and the local market, and who can explain a valuation face to face. The opportunity is to use AI for the parts of the job that do not require that judgement — drafting a first description, preparing comparable evidence, cutting time-to-post, producing a video walkthrough from photographs — and to spend the time saved on the parts that do: honest pricing conversations, accurate condition reporting, and viewings that answer real questions.

A practical routine for an agent might look like this. First, gather comparables before you draft anything, and note the adjustments you are making for condition, location, and tenure. Second, use any drafting assistance available to produce a first version, then edit it yourself for accuracy, removing any claim you cannot verify and adding the local detail only you know. Third, disclose any material AI alteration, briefly, in the listing or in your own marketing notes. Fourth, prepare for assisted buyers: expect enquiries that reference specific comparisons or price points, and answer them with your evidence rather than with general reassurance. Fifth, track what happens — which listings complete fastest, which price guidance proved accurate, which descriptions generated the best-qualified viewings — and use that record to refine the next listing. None of this requires a new platform. It requires treating AI as a junior colleague whose work you check, not as a replacement whose work you publish unchecked.

A playbook for car dealers

For car dealers, the Motors results OLX cites are the most directly relevant part of the announcement. More than 70% of dealers accepting AI-made videos, and agentic search live across Motors, suggest a category where visual presentation and specification matching are already being reshaped by AI. A dealer who still relies on a handful of hurried photographs and a copied specification sheet will look increasingly thin next to a dealer whose listings offer a clear video, complete history and condition notes, and a price the dealer can justify from comparable stock.

The dealer playbook follows the same logic as the agent playbook, with a different emphasis. Condition disclosure matters even more for vehicles, because a buyer who travels to see a car and finds undisclosed damage will not simply walk away; they will tell others. Video helps here, provided it shows the car as it is, including wear, rather than presenting a polished impression that the viewing contradicts. Pricing guidance helps here too, provided the dealer checks it against their own stock costs, reconditioning spend, and local market. And speed matters: a buyer alerted by an assistant about a newly listed vehicle in their specification and price range is a buyer who may contact several dealers at once. The dealer who replies first with availability, a viewing time, and a straight answer about history and condition has a real advantage, and it has nothing to do with technology. It is simply good selling, accelerated by a better-matched enquiry.

A playbook for independent retailers

Independent retailers selling through marketplaces — whether that means second-hand goods, refurbished electronics, furniture, vintage items, or specialist equipment — face a slightly different version of the same change. Their listings are often one-offs, so pricing guidance based on comparables is both more valuable and more fragile. Two similar-looking items can differ enormously in condition, provenance, or completeness, and an AI that has not seen those differences can mislead. The retailer’s defence is detail: photographs of flaws as well as features, a description that says what is included and what is not, an honest condition grade in plain language, and a price note that explains, briefly, why this item sits where it does.

For retailers, the buyer-assistant trend also rewards a habit many small shops already have but do not always carry online: knowing your stock well enough to answer a specific question quickly. When a buyer asks, through an assistant or directly, whether an item will fit a particular space, work with a particular model, or arrive by a particular date, a precise answer closes sales. A vague answer sends the buyer back to the assistant for another option. Building a small internal habit — keeping measurements, compatibility notes, and delivery options with each listing, rather than in someone’s head — pays off disproportionately once assistants start doing the searching for buyers.

A playbook for trades and service businesses

Tradespeople and service businesses may read a marketplace announcement and assume it does not apply to them. It does, by analogy, because the same buyer behaviour is spreading to service marketplaces and messaging channels: customers describe a job in ordinary language, sometimes with a photograph or a voice note, and expect a matched, priced, and available response. The lessons transfer directly. First, make your offer easy for a system and a human to match: clear service descriptions, the areas you cover, the types of job you do and do not take, and realistic availability. Second, price transparently: where a fixed price is possible, state it; where it depends on inspection, say what affects it and what a typical range looks like. Third, disclose AI use where it matters: if a quotation, an image, or a description was AI-assisted, a brief note preserves trust. Fourth, respond specifically and quickly, and keep the conversation in the channel the customer chose. Fifth, keep the final decision — the quote, the scope, the start date — visibly yours. Customers who feel they are dealing with a person, not a system, are more likely to accept a price, wait for a slot, and recommend you afterwards.

Measurement: how to tell whether agent-assisted selling is helping you

General advice, offered here without any claim about a particular platform, is to measure the change in the same way you would measure any sales change: before and after, on the numbers that actually matter. Track posting completion — how many listings you start versus how many go live — because that is where drafting assistance should help first. Track time-to-post, from deciding to sell to a live listing, because that is where automation should save hours. Track enquiry quality, not just enquiry volume: how many enquiries lead to a viewing, a call, or a sale, rather than how many messages arrive. Track price accuracy: how often your asking price survives negotiation unchanged, and how often you reduce it, and by how much. And track disclosure complaints or surprises — viewings where the buyer expected something different — because a rise there is an early warning that assistance is drifting into misrepresentation.

If you adopt any AI drafting or pricing help, run a small, honest comparison for a few weeks. Publish some listings with assistance and some without, keep the category and price range similar, and compare completion, time, enquiry quality, and final sale price. You do not need sophisticated analytics for this; a simple spreadsheet and a disciplined note of what you changed will tell you more than a vendor’s headline figure. The point is not to prove that AI works in general. It is to learn whether it works for your stock, your customers, and your market, and where you still need to intervene.

Honest limits / what to watch

The first limit is the most obvious and the most important: none of this has been proven at UK scale, on UK platforms, with UK buyers and sellers. Agentic Selling is live for a fifth of private sellers on one Polish property platform. DealFinder’s beta involved 2,000 users in Portugal. Those are real, live results, and they are encouraging, but they are also narrow. A tool that works for private sellers on Otodom may behave differently for UK estate agents, for car dealers with large stock, or for retailers with one-off items. A buyer assistant that converts well in a Portuguese beta may face different expectations, price sensitivity, and messaging habits elsewhere. Treat the OLX figures as evidence that the approach is promising, not as a promise that it will work identically for you.

The second limit concerns pricing guidance itself. Data-backed guidance is only as good as the data behind it, and marketplace data has blind spots. It may underweight condition differences that photographs do not capture, over-weight recent outliers, or miss local factors — a new development nearby, a change in transport links, a seasonal effect — that a local seller or agent would know immediately. Guidance that looks precise can therefore be misleading in exactly the cases where precision matters most: unusual items, thin markets, and fast-moving categories. The safeguard, which OLX has built into its design, is human approval. The risk is that approval becomes a habit of clicking through. If sellers stop checking guidance because it is usually right, the occasional wrong figure will slip through, and it will slip through on the listing where it costs the most. The discipline of checking every price, even when the system has been right ten times in a row, is tedious — and it is the whole point.

The third limit is disclosure, and it cuts both ways. Disclosing AI-altered content builds trust, but disclosure alone does not fix a misleading listing. A photograph that has been enhanced to remove a flaw, disclosed in small print, is still a photograph that misleads a buyer who does not read the note. A description drafted by AI and checked carelessly can still contain a claim the seller cannot support. The standard that matters is not whether AI use was mentioned, but whether the listing is accurate. Sellers should treat disclosure as a minimum, not as a shield, and should hold AI-assisted content to the same accuracy standard they would hold their own words to — because, in the eyes of the buyer, it is their words.

The fourth limit is the buyer side. Conversational assistants make searching easier, but they also insert a new intermediary between buyer and seller. The buyer may trust the assistant’s shortlist more than they trust any individual listing, which means a seller whose listing is poorly structured, incompletely specified, or slow to answer may never reach the buyer at all, even if the item itself is ideal. That is a new form of invisibility, and it rewards the same unglamorous work — complete data, clear photographs, honest descriptions, fast replies — that good marketplace selling has always rewarded, only now with less room for error. It also raises a question sellers should keep asking: what does the assistant optimise for, and does that match what your buyer actually values? An assistant that prioritises price will surface the cheapest option, not necessarily the best-conditioned one. A seller competing on quality needs to make that quality visible in the data the assistant can read, not just in a viewing the buyer may never reach.

The fifth limit is the survey evidence itself, and it deserves a careful reading. The jobs survey covered young people in larger cities in five countries; it does not describe all workers, all ages, or all places. The real estate survey, as OLX itself notes, is not representative of all home seekers. The headline figures — 92% to 96% personal AI use, fewer than 41% feeling ready for labour-market change, 5% trusting an AI valuation over a human agent’s — are useful signals, not universal truths. They suggest a gap between use and trust, and a preference for human judgement in high-stakes decisions, but they do not tell you what your own customers think. The only reliable way to learn that is to ask them, watch how they behave, and notice where they hesitate. If buyers in your market keep asking for a human valuation, a viewing, or a phone call before committing, believe them, even if the industry narrative says everything is moving to full automation.

Finally, there is a broader watch item: agent commerce is still being defined, and the rules — around disclosure, liability, data use, and what an agent is allowed to commit a buyer or seller to — are still catching up. A tool that drafts is low-risk. A tool that negotiates, commits, or pays on your behalf is a different proposition entirely, and it should be adopted with far more caution, clear limits, and a record of what was agreed. OLX’s current tools, as described, stay on the safer side of that line: they guide, draft, search, and alert, while people decide. That restraint is a feature, not a lack of ambition. UK sellers should look for the same restraint in any agent tool they adopt, and should be wary of any tool that asks them to surrender pricing, disclosure, or commitment decisions without a clear, checkable explanation of why it chose what it chose.

None of these limits diminishes the launch. If anything, they clarify why it matters. OLX has not announced a machine that sells for you. It has announced assistance that helps you sell better, in a form that respects the buyer’s desire to understand and the seller’s responsibility to decide. That is a more modest claim than much of the agent-commerce hype, and a more useful one. The businesses that benefit most will not be those that adopt every AI feature first, but those that adopt the right assistance, check it properly, disclose it honestly, and keep the human relationship — the viewing, the conversation, the straight answer — at the centre of the sale.

FAQs

What is OLX Agentic Selling?

Agentic Selling is OLX Group’s seller-side AI tool, unveiled at CLAIM Online 2026 on 27 October 2026. It guides private sellers from valuation to posting, offering transparent, data-backed pricing guidance and listing drafts, while final pricing and disclosure decisions stay with the seller. It is currently live for 20% of private sellers on Otodom in Poland, which suggests a phased rollout rather than an immediate launch across all of OLX’s markets.

What is DealFinder and how does it work?

DealFinder is OLX’s buyer-side assistant on WhatsApp. A buyer can search by text, image, or voice, and DealFinder finds items, sets price and item alerts, and notifies the buyer when a matching item appears or a price moves. It is being scaled after a beta in Portugal with 2,000 users, in which it achieved an 18% conversion rate to contact. In practical terms, it shifts the buyer from repeatedly searching a marketplace to being monitored and alerted, which tends to produce warmer, better-matched enquiries.

Can UK small businesses use Agentic Selling or DealFinder now?

Not directly, on the basis of the information published with the launch. OLX serves Europe and South Africa, and the tools described are live or in beta in Poland and Portugal respectively. For a UK business, the immediate value is as a guide to where marketplace selling is heading: guided pricing, drafted listings, conversational buyer assistance on messaging apps, and disclosure of AI-altered content are practices worth adopting now, on the platforms you already use, rather than features to wait for.

Why does OLX emphasise trust and human control so strongly?

Because its own October 2026 research suggests customers use AI widely but do not yet trust it with high-stakes decisions. In the jobs survey, 92% to 96% of young urban respondents used AI tools personally, yet fewer than 41% felt ready for labour-market changes. In the real estate survey, only 5% would trust an AI valuation over a human agent’s estimate, and 78% wanted AI-altered listings disclosed. OLX’s leaders have argued, in paraphrase, that AI should handle complexity while people keep the decisions that matter, and that showing why a tool recommends something is part of earning trust in its outcome.

What should a UK seller do differently after this launch?

Start with the basics that agent-assisted marketplaces reward: complete and accurate listings, photographs that show condition honestly, pricing you can explain from comparable evidence, brief disclosure where AI materially altered a listing, and fast, specific replies to enquiries. Measure the results — posting completion, time-to-post, enquiry quality, and price accuracy — and keep checking any AI draft or price suggestion yourself before publishing. The tools will change, but the discipline of honest, checkable selling is what will keep working whichever platform your customers use next.

CTA: Get a Free AI Consultation

Sources: OLX Group press release and Unite.AI coverage of the CLAIM summit launch.