AI · Review
ActiveMoss Review 2026: AI Support Inside Web Products
Moss is an AI support agent embedded in web products that answers from company documentation, guides users through live screens and proposes improvements for team approval. Its hands-on product workflow is clearly described, but pricing and independent security details are not published.
By the DappsAtlas editorial team · Updated
DappsAtlas score
7.2/10
Good
Our verdict on Moss
7.2/10Moss is aimed at software businesses that want customer support and onboarding to happen inside their product. Its main strength is the combination of documentation-based answers, live walkthroughs, controlled actions and session analysis, with staff approval for proposed changes. The main caveats are unpublished pricing and limited independently verifiable security detail: Moss reports a completed penetration test, but its SOC 2 and ISO 27001 certifications are still in progress. Teams should check data controls, integrations and deployment terms before adopting it.
Score breakdown
How Moss scores
- Security
- 5.0/10
- Ease of use
- 8.0/10
- Fees
- n/a
- Features
- 8.0/10
Scores run from 0 to 10 and are based only on facts we could verify. Where information was missing we mark the score n/a rather than guess.
What is Moss?
Moss is an AI support agent designed to run inside a company’s web product. It answers questions using connected documentation, guides customers through the live interface and can perform configured actions, such as filing a support ticket. Its monitoring tools review sessions against criteria set by the team and highlight problems for follow-up.
The product is aimed at businesses that want to support onboarding and customer questions inside their software rather than relying only on static tutorials or an external help queue. Moss describes one-line SDK installation and offers cloud, customer-cloud and on-premises deployment. Pricing is not published, and SOC 2 and ISO 27001 are still in progress.
How Moss works
Moss is embedded in a web product using an SDK. It reads the live page at runtime, uses connected documentation to answer questions with citations, and can point out interface controls or guide users through a task. Depending on the permissions a team sets, it can also take actions such as filling a form or filing a support ticket. Teams can designate elements it must not interact with and define when it should hand a conversation to a person.
After each interaction, Moss can evaluate the session against criteria written by the product team. It groups recurring problems and prepares evidence-backed changes for staff to review. Moss says proposed changes do not publish silently: designated team members can approve, edit or dismiss them, and published versions can be restored.
5 features
Key features
01
On-screen support
Answers user questions on the current product screen and cites connected documentation as sources.
02
AI walkthroughs
Guides users through live interface steps, with walkthroughs authored from documentation or generated when a flow is not covered.
03
Agent procedures
Runs plain-English procedures to check systems, take permitted actions and hand off to staff when configured.
04
Session monitoring
Scores sessions against team-defined criteria and attaches reasoning and session evidence to flagged failures.
05
Evidence-led improvements
Groups recurring issues and prepares proposed product or support changes for a designated person to approve, edit or dismiss.
Pros and cons
What we like
- Combines in-product answers, walkthroughs and permitted actions in one support agent.
- Uses connected documentation as the basis for cited answers.
- Lets teams test procedures in a sandbox and set limits on what the agent can do.
- Provides session review and evidence for proposed improvements, with human approval before changes publish.
- Offers customer-cloud and on-premises deployment options, including air-gapped environments.
Watch out for
- Pricing and subscription terms are not published.
- SOC 2 and ISO 27001 certifications are still in progress.
- The penetration-test report and details of any independent security audits are not published.
- Using live screen context and customer data creates privacy and access-control risks, even with redaction and purge options.
- The product depends on AI interpreting the interface and following procedures correctly; safeguards reduce but do not remove the chance of errors.
Moss fees
Moss does not publish pricing or subscription fees on its homepage. Contact the company for a quote and confirm any implementation or usage charges before signing up.
Is Moss safe?
Moss says customer data is not used to train its models, redaction is enabled by default, and screen data can be purged during processing with an audit log. It names OpenAI, Anthropic, Google AI, OpenRouter and Baseten as model subprocessors. These are company-stated controls; customers should review data-processing terms, subprocessors, retention settings and access permissions for their chosen deployment.
Moss says a penetration test was completed in May 2026, while SOC 2 and ISO 27001 are in progress. It does not publish the penetration-test report or details of independent security audits. This is a hosted software service, not a crypto protocol: there are no stated tokens, smart contracts or on-chain custody risks. No specific regulatory status is published.
6 steps
How to get started with Moss
- 1
Arrange a demo or ask Moss to assess a demo environment before adding code.
- 2
Choose a deployment model: Moss cloud, your own cloud or on-premises.
- 3
Add the SDK line to your web product.
- 4
Connect the help center, product documentation or uploaded files Moss should use for answers.
- 5
Define procedures, permissions, handoff rules and session-review criteria.
- 6
Test procedures in the sandbox, then review and approve them before publishing.
Who Moss is for
Moss may suit B2B software teams with recurring onboarding or support questions, especially those that want users guided in the interface and want session-level evidence to help improve support. It is less suited to teams that need published pricing, completed security certifications or a fully self-service product evaluation.
Moss FAQs
What is Moss AI?
Moss is an AI support product for web applications, not a cryptocurrency or blockchain service. The product embeds an agent in a company’s interface to answer questions, guide users and carry out permitted tasks.
How does Moss work?
Moss says it reads the user’s screen at runtime and answers using connected help documentation, with source citations. It can guide a user through the interface or perform configured actions, then flag session failures and suggest changes for staff review.
How does Moss handle customer data?
Moss says it can run in its own cloud, a customer’s cloud or fully on-premises, including air-gapped environments. The company says customer data is not used to train models and that redaction is enabled by default; prospective customers should confirm the deployment and data-handling terms that apply to them.
How much does Moss cost?
Moss does not publish pricing on its homepage. Ask the company for a quote and confirm any subscription, implementation or usage charges before deployment.
How do you get started with Moss?
The homepage describes an SDK-based setup, connecting a help center, documentation or uploaded files, and configuring procedures and review criteria. Teams can dry-run procedures in a sandbox before publishing them.
What integrations does Moss support?
Moss says it files Zendesk tickets and reads knowledge through Intercom and Document360, with additional connectors expanding. Confirm whether the integrations your team needs are supported before adopting it.
Compare
Moss alternatives
OpenArt is a subscription AI studio for generating images, video and music with 100+ models, priced by monthly credits, aimed at creators and marketers.
Venice is a privacy-focused AI platform for chat, images, video and API access, with a VVV token on Base, aimed at creators and developers.
Kismet is a beta AI agent for Unreal Engine developers that edits project assets and checks changes with editor and compiler feedback.
Talking Tree is a subscription legal AI and document platform for startups and small businesses.
Salad (SaladCloud) is a distributed GPU cloud that runs AI containers and inference on idle consumer GPUs, aimed at cost-sensitive developers and teams.
Huralya’s Hura is a mental wellness app for people seeking AI coaching, mood tracking, journaling and everyday self-care tools.
Same category
More AI dApps
ZyFAI is a self-custodial DeFi yield agent that automates allocation across vetted lending pools on Base, Arbitrum, Ethereum, Sonic and Plasma.
CasperPractice is an AI-assisted CASPer test-prep platform for applicants who want timed practice questions and feedback on their answers.
Yogi is an AI consumer-insights platform for brands that want to analyze feedback across reviews, support, social channels, and surveys.
Handsy is a desktop webcam app that uses local AI to alert users when it detects habits such as nail biting or face touching.
BeeOS is infrastructure for developers running AI agents across persistent sandboxes, computers, phones, and cloud environments.
QuantiX is an AI-assisted strategy research and automated trading platform for retail traders using their own exchange accounts.
How we checked
Sources
- 1.openmoss.ai/MOSS-Image/
- 2.mossmotors.com
- 3.plantglossary.com/types-of-mosses/
- 4.moss.com
- 5.bhg.com/how-to-grow-moss-11941532
- 6.britannica.com/plant/moss-plant
Last updated . Spotted an error? Tell us and we will correct it.