Appen
APXAustralian training-data company that builds and validates the datasets frontier models are trained on, through its AI Data Platform (ADAP) and product lines spanning frontier alignment, agentic AI, speech and audio, multimodal, physical AI and model integrity.
Products
Selected offerings — the AI products recorded here, not the vendor's full catalogue.
-
AI Data Platform Platform
Appen's data platform (ADAP), where automation and human oversight run together to produce expert-validated training data across six lines: frontier alignment, agentic AI, speech and audio, multimodal, physical AI and model integrity.
-
MatrixGo Platform
MatrixGo is Appen China's enterprise data annotation platform, offering image, point-cloud, video, text and speech annotation tools with project management, configurable workflows and AI-assisted pre-labelling, sold as public-cloud SaaS or private-cloud on-premises deployment.
-
Appen's licensable catalogue of pre-collected, pre-licensed AI training datasets: 597 datasets across nine categories, from reinforcement-learning tasks and verifiers to speech, code and enterprise data, under a perpetual, non-exclusive commercial training licence.
-
Training and evaluation data services for AI agents: task/verifier design, trajectory and failure-mode analysis, golden-trajectory creation, RL environment design, RAG evaluation, and SWE-driven deep evaluation -- verified live with six named sub-services.
-
Expert human data for aligning frontier models: chain-of-thought reasoning traces, subject-matter-expert RLHF preference ranking from verified PhDs, MDs and JDs, supervised fine-tuning demonstrations, adversarial red teaming, evaluation rubric design and a multilingual LLM-as-a-judge managed service.
-
Human evaluation services for model trust and compliance: hallucination and factuality benchmarking, EU AI Act and NIST AI RMF audits, A/B production arena testing, bias detection with remediation datasets, continuous post-deployment monitoring and LLM-as-a-judge rubric design.
-
Sensor-synchronized training data for robots and embodied systems: egocentric first-person capture with wrist pose and gripper state, robotics task demonstrations, sensor-fusion sequences, LiDAR annotation, autonomous-vehicle data and world-model data.
Closest alternatives 1
Similar products 119
Closest alternatives are pairs a vendor page itself compares. Similar products are matched from what each product does; nothing in that group is asserted as a rivalry.
Sources
- Total funding
-
Checked — none published
Listed on the ASX; equity funding history is not the relevant figure for a public company and none is published as a total.
- Market cap
-
Market capitalisation is A$290.01M (2026-09-04) — quoted in Australian dollars. This column stores raw USD, and converting would mean choosing an exchange rate this record cannot cite, so it is null rather than a self-made conversion.
- Founded
-
The infobox gives 1996, but it is not corroborated on Appen's own site and this record does not carry a founding year on one aggregated infobox line alone.
- Revenue
-
A$361.92M trailing twelve months (2026-09-04), again in Australian dollars — same conversion problem as valuation. Wikipedia's US$273.0M is a 2023 figure and three years stale.
Something wrong here? Send a correction — quote this company id: appen.