Online classifieds platform with AI-assisted moderation
Netbase delivered a bilingual classifieds marketplace, in English and Nepali, with paid ads, search, messaging, payments and an AI filter that flags offensive content into an administrator review queue. The client is not named, and no traffic, accuracy or commercial result is claimed.
Submit a project See the related service
Record at a glance
- Evidence level
- Portfolio record: delivered scope from the project documents, attested by Netbase's CEO
- Client
- A founder launching a classifieds marketplace; not named
- Interface
- English and Nepali
- Service
- AI Workflow Automation, with Custom Product Engineering scope
- Delivery
- Six milestones over four months, then six months of support
- Published metrics
- None
Context
A founder planned to launch an online classifieds marketplace in two languages, English and Nepali. Sellers would post paid ads, buyers would search by category and location, and the two sides would talk and rate each other on the platform. Because anyone could post, the founder also needed a way to catch offensive content before it damaged trust in the marketplace, without hiring a large moderation team from day one.
Netbase role
Netbase delivered the project end to end: consultancy and requirements, wireframes and interface design, backend and API development, deployment, training and post-launch support. The work was planned in six milestones over four months, followed by six months of support.
Delivered scope
- requirements, wireframes and interface design for buyer, seller and administrator journeys;
- a Laravel backend exposing REST APIs;
- OTP-verified user accounts and paid ad posting;
- category and location search;
- buyer and seller ratings and on-platform messaging;
- event ticket ads, a blog and a forum;
- two online payment gateway integrations;
- multi-language SEO for the two interface languages;
- deployment on AWS, administrator training and six months of support.
The AI-assisted moderation workflow
The moderation component used AI-powered content filtering to detect offensive content and flag it into an administrator moderation dashboard. The design choice matters: the filter did not silently delete posts. It raised items for a person to review, which kept the final decision with the marketplace operator and gave a place to handle mistakes in either direction.
This is the pattern the AI Workflow Automation service is built around: an automated first pass, a defined review point and a human decision where the impact is high.
Evidence level and limitations
- The record describes scope from the proposal, agreement and requirements documents. It does not measure how well the moderation performed.
- The AI moderation service used is not recorded in the documents, so no vendor is named.
- Capacity and response-time targets in the requirements were design goals, not measured results, and are not reported here.
- The client is an individual and is not named; the target country is not stated in the documents.
- No screenshots are shown. The image on this page is illustrative.
Lessons for a similar project
If you are planning a marketplace with user-generated content, decide these before choosing a technical approach:
- the moderation policy: what counts as offensive or prohibited, in each language you support;
- the review queue: who reviews flagged items, how quickly, and with what tools;
- false positives and appeals: how a wrongly flagged seller gets a fair answer;
- data rights: whether user content can be sent to an external AI service, and under what terms;
- an evaluation set of real or realistic posts to test the filter before launch.
The product work around the filter is usually larger than the filter itself. Accounts, payments, search and messaging are Custom Product Engineering scope, planned as milestones in the same project. The full operating workflow, from listing intake to a closed deal, is set out in marketplace and classifieds operations, and the sector view is on the retail commerce and marketplaces page.
Questions buyers ask
Did the AI remove posts on its own? No. The filter flagged offensive content into an administrator moderation dashboard, and a person made the decision. Nothing was deleted silently.
Which AI moderation service was used? The project documents do not record it, so this record names none and makes no vendor claim.
How long did delivery take? Six milestones over four months, followed by administrator training and six months of support.
Planning a marketplace with moderation?
See the other portfolio records, how evidence is labelled on the methodology page, and how milestones run on the project delivery page. When you are ready, submit a project brief. OutsourcingVN is operated by Netbase JSC and is Netbase's own outsourcing-services platform.