Arwen

Arwen is a UK-based company that provides a real-time platform for managing the comment layer of social media. The platform connects to social networks through secure APIs and collects comments across both paid and organic content, then classifies each comment for intent, sentiment, topic and risk before taking automated or human-guided action. Four products sit on this platform: Moderate (automatic removal of toxic, abusive and spam comments across organic posts and ad formats), Engage (surfacing buyer intent, questions, complaints and praise, with suggested on-brand replies), Protect (detection, investigation and escalation of threats and emerging risks) and Signals (narrative, role-specific recommendations and trend detection intended to replace dashboard-style reporting).

The classification layer combines proprietary classifiers, fine-tuned large language models, open-source techniques and rule-based logic, with the company stating it applies LLMs where context and interpretation matter and rules where precision is more reliable. The platform handles text and emoji by default, supports over 30 languages, and can be extended to images, GIFs and video. Configuration is per client — policies, thresholds, tone and escalation rules are set by the customer — and the company emphasises transparency of every automated action, human review of ambiguous or high-risk cases, and continuous tuning to reduce false positives and over-moderation. Coverage extends across organic posts, paid ads, cross-platform ads, branded partnership ads, Spark and non-Spark ads, and promoted videos.

Arwen positions the platform for enterprise scale, supporting deployments from a handful of profiles to global operations, and markets it to paid social, community management, customer service, social listening, Trust & Safety, risk and reputation, and social leadership functions.

Founding story

Arwen was founded in 2020 by a team whose stated experience spans AI, product, service, security and enterprise software. CEO Matt McGrory says he started the company after years of building technology for brands and observing teams struggle with noise and risk in social comment sections, and that the racist abuse during the 2020 European Championship prompted him to act. Dr David Cole is listed as CTO. The founders' stated aim was to help teams make calmer, better decisions under pressure by supporting rather than replacing human judgement.

Business model

Arwen sells a software platform, complemented by services such as managed community management and high-touch consultative support, to organisations that operate social media accounts and advertising at scale. Deployment is via no-code connection to clients' social profiles, with filters, policies, escalation rules, reporting and team management configured per client; moderation is delivered in-house using a combination of AI, keyword filtering and manual review rather than third-party contractors.

Arwen sells its platform to enterprise customers on a business-to-business basis, with products (Moderate, Engage, Protect, Signals) and services such as community management; the company frames its value proposition in ROI terms, stating clients see $4 to $5 returned for every $1 spent through time saved, costs avoided and revenue gained.

Traction

The company states it manages over 750 million followers globally and reports customer-cited outcomes including 94% less time spent on moderation and engagement, sub-30-second comment responses, up to 19% reduction in customer acquisition cost, and 4-5x ROI. Named customers and references include Hinge, Oanda, The Daily Wire and Northampton Saints, with partnerships announced covering the ATP (300 tennis players) and the DOSB.

Latest developments

Recent company announcements include a partnership with the ATP to launch a Safe Sport programme covering 300 tennis players and a partnership with the German Olympic Sports Confederation (DOSB) to protect German Olympic athletes online, alongside the publication of a buyer's guide to AI auto-moderation services.