Manus AI

Name: Manus AI
Brand: Manus AI
Rating: 1.6 (68 reviews)

AI NativeLeader

Manus from Meta is the general AI agent that turns your thoughts into actions

Est 2022|Valuation $2B|Raised $85M|34 Ref Customers

LinkedIn 127K|Web traffic 34.3M|FTEs 171|HQ Singapore, SG

MarketsSuper Agents

Published Oct 25, 2026

01Common complaints

1. If enterprise-grade compliance evidence (SOC 2, ISO 27001, a formal bug bounty) is required and cannot be produced during diligence. 2. If billing or cancellation risk is unacceptable given reported customer complaints. 3. If reliable agent fidelity is required (i.e., the agent must not claim actions it did not perform). 4. If deploying production websites with strict uptime requirements—some reports indicate deployment instability and credit waste. 5. If broad payment-provider flexibility is required—some user reports reference limited payment options (verify current status with the vendor).

Deal disqualifiers (poor fit)

Regulated environments requiring SOC 2/ISO attestation and strong contractual guarantees.
Mission-critical production deployments where downtime or rollbacks are unacceptable.
Teams that require highly predictable costs and straightforward cancellation and billing controls.

Limitations documented by the vendor

Credits reset on subscription end: unused monthly credits are forfeited after subscription expiration per Help Center documentation.
Refund constraints: The Help Center indicates that canceling generally prevents future renewals but does not guarantee refunds for the current period; refund eligibility depends on terms and circumstances.

Limitations reported by users (to verify in pilot)

Task crashes and frozen sandbox sessions.
Deployment reliability issues.
Difficulty with support escalations.
Potential security exposure allegations (community claims, unverified).

Why customers leave / churn reasons (common patterns)

Billing and cancellation friction leading to trust breakdown.
Task unreliability (crashes, loops, inconsistent outputs).
Project loss or rollbacks.
Deployment and hosting limitations (payment rails, hosting constraints).

Where it struggles (based on user reports)

Reported phantom actions where steps are claimed but not reflected in outputs.
Web development tasks with correctness issues (HTML/JS mismatches reported by users).
Long-running tasks that may crash or lose context.

02Why they buy

Distinguishing features supported by evidence (partial)

Integrated agent experience: Agent Mode with tool use and artifact creation; widely described in press and community discussions.
Press-covered adoption metrics: Reports of rapid growth and reported ARR have been noted in multiple outlets; these are third-party reports and should be validated in procurement.

Claims that require cautious treatment

Benchmark leadership and reported success-rate improvements are vendor-provided and not independently validated in the materials reviewed.

03Field reports

Review platform availability note (G2/Capterra)

Searches for G2 and Capterra did not surface substantial listing pages or review volumes for Manus AI / manus.im in the materials reviewed; Trustpilot was the most visible consumer review source identified.

Trustpilot snapshot (negative skew observed)

Trustpilot shows a low average rating with numerous complaints focused on billing, cancellation, refunds, reliability, and task fidelity. Representative complaint themes: - Reports that the agent claimed to complete tasks without observable results. - Customers describing access problems after payment and difficulty obtaining refunds. - Statements that canceling subscriptions was difficult or did not behave as expected. - Accounts of lost work and limited or slow support responses. Balanced note: Review sites can skew negative because dissatisfied customers are more likely to submit reviews. However, the consistency of themes (cancellation difficulty, unexpected charges, unreliable execution, missing refunds/support) across multiple reviews is a signal that warrants attention in diligence.

Community/social testimonials (Reddit / Hacker News)

Community threads include user reports describing unperformed changes after agent runs and accounts of users migrating away from the product to alternative assistants.

Hacker News discussions focus on skepticism about product architecture and benchmark claims.

Why customers choose Manus (inferred from reviews/press)

Vendor claims of autonomous artifact creation (websites, apps, spreadsheets).
Ability to run tasks in a virtual computer environment.
Press coverage and reported metrics that indicate notable market attention.

Availability note

This research pass did not find robust, third-party-verified customer case studies hosted on the vendor site comparable to established B2B vendors' case-study libraries.

Vendor-published performance claims (not independently validated)

Vendor release communications include claims of improved task success rates and higher user satisfaction in internal testing.

Critical evaluation:

The vendor does not disclose full methodology, sample sizes, task lists, or benchmarking protocols in the materials accessed. These should be treated as vendor-provided claims until validated by third-party tests or internal pilot programs with reproducible metrics.

04Best fit

Best-fit segments (pragmatic)

Individuals and small teams that can tolerate instability and oversee outputs.
Use cases where failures are low-stakes (prototypes, drafts, exploratory research, internal tooling).

Personas

Founders or product managers experimenting with quick prototypes.
Growth or marketing operators generating assets and research briefs.
Analysts performing competitive research with verification steps.
Developers using an agent as scaffolding or a productivity aid.

Qualification criteria (good fit)

High tolerance for iteration and quality assurance.
Tasks that can be validated quickly.
Active management of credits and usage.

05The pitch

What the product does (objective value proposition)

Manus AI is a web/app-based agentic AI product implemented as an autonomous execution layer rather than a traditional chat assistant. It is designed to attempt multi-step tasks end-to-end (for example, research, coding, data analysis, and website/app building) by planning work, using integrated tools (browser, code execution, files), iterating, and producing artifacts (apps, spreadsheets, reports). Manus communicates improvements in Agent Mode performance via vendor messaging and offers capability bundles such as wide research, website building/hosting, mobile development, and a visual Design View. Important caution: A consistent thread across community discussions and consumer reviews is a gap between the product's autonomous execution claims and observed task completion—users report instances where the agent reported progress or completion without verifiable changes, particularly for web/app tasks.

Target market and positioning

Primary target: Individual power users, indie builders, marketers, founders, and small teams seeking agentic assistants that execute multi-step workflows (research and build).
Secondary target: SMB teams needing shared credits and concurrent task execution (Team plan).
Positioning: A general-purpose AI agent designed to execute workflows in a sandbox/virtual computer environment rather than solely answering questions.

Get a comprehensive analysis of Manus AI including market position, competitive landscape, adoption trends, and peer benchmarks.

Download full report →

Published Oct 25, 2026

01Common complaints

If enterprise-grade compliance evidence (SOC 2, ISO 27001, a formal bug bounty) is required and cannot be produced during diligence.
If billing or cancellation risk is unacceptable given reported customer complaints.
If reliable agent fidelity is required (i.e., the agent must not claim actions it did not perform).
If deploying production websites with strict uptime requirements—some reports indicate deployment instability and credit waste.
If broad payment-provider flexibility is required—some user reports reference limited payment options (verify current status with the vendor).

Deal disqualifiers (poor fit)

Regulated environments requiring SOC 2/ISO attestation and strong contractual guarantees.
Mission-critical production deployments where downtime or rollbacks are unacceptable.
Teams that require highly predictable costs and straightforward cancellation and billing controls.

Limitations documented by the vendor

Credits reset on subscription end: unused monthly credits are forfeited after subscription expiration per Help Center documentation.
Refund constraints: The Help Center indicates that canceling generally prevents future renewals but does not guarantee refunds for the current period; refund eligibility depends on terms and circumstances.

Limitations reported by users (to verify in pilot)

Task crashes and frozen sandbox sessions.
Deployment reliability issues.
Difficulty with support escalations.
Potential security exposure allegations (community claims, unverified).

Why customers leave / churn reasons (common patterns)

Billing and cancellation friction leading to trust breakdown.
Task unreliability (crashes, loops, inconsistent outputs).
Project loss or rollbacks.
Deployment and hosting limitations (payment rails, hosting constraints).

Where it struggles (based on user reports)

Reported phantom actions where steps are claimed but not reflected in outputs.
Web development tasks with correctness issues (HTML/JS mismatches reported by users).
Long-running tasks that may crash or lose context.

02Why they buy

Distinguishing features supported by evidence (partial)

Integrated agent experience: Agent Mode with tool use and artifact creation; widely described in press and community discussions.
Press-covered adoption metrics: Reports of rapid growth and reported ARR have been noted in multiple outlets; these are third-party reports and should be validated in procurement.

Claims that require cautious treatment

Benchmark leadership and reported success-rate improvements are vendor-provided and not independently validated in the materials reviewed.

03Field reports

Review platform availability note (G2/Capterra)

Trustpilot snapshot (negative skew observed)

Trustpilot shows a low average rating with numerous complaints focused on billing, cancellation, refunds, reliability, and task fidelity.

Representative complaint themes:

Reports that the agent claimed to complete tasks without observable results.
Customers describing access problems after payment and difficulty obtaining refunds.
Statements that canceling subscriptions was difficult or did not behave as expected.
Accounts of lost work and limited or slow support responses.

Balanced note: Review sites can skew negative because dissatisfied customers are more likely to submit reviews. However, the consistency of themes (cancellation difficulty, unexpected charges, unreliable execution, missing refunds/support) across multiple reviews is a signal that warrants attention in diligence.

Community/social testimonials (Reddit / Hacker News)

Community threads include user reports describing unperformed changes after agent runs and accounts of users migrating away from the product to alternative assistants.

Hacker News discussions focus on skepticism about product architecture and benchmark claims.

Why customers choose Manus (inferred from reviews/press)

Vendor claims of autonomous artifact creation (websites, apps, spreadsheets).
Ability to run tasks in a virtual computer environment.
Press coverage and reported metrics that indicate notable market attention.

Availability note

This research pass did not find robust, third-party-verified customer case studies hosted on the vendor site comparable to established B2B vendors' case-study libraries.

Vendor-published performance claims (not independently validated)

Vendor release communications include claims of improved task success rates and higher user satisfaction in internal testing.

Critical evaluation:

The vendor does not disclose full methodology, sample sizes, task lists, or benchmarking protocols in the materials accessed. These should be treated as vendor-provided claims until validated by third-party tests or internal pilot programs with reproducible metrics.

04Best fit

Best-fit segments (pragmatic)

Individuals and small teams that can tolerate instability and oversee outputs.
Use cases where failures are low-stakes (prototypes, drafts, exploratory research, internal tooling).

Personas

Founders or product managers experimenting with quick prototypes.
Growth or marketing operators generating assets and research briefs.
Analysts performing competitive research with verification steps.
Developers using an agent as scaffolding or a productivity aid.

Qualification criteria (good fit)

High tolerance for iteration and quality assurance.
Tasks that can be validated quickly.
Active management of credits and usage.

05The pitch

What the product does (objective value proposition)

Manus communicates improvements in Agent Mode performance via vendor messaging and offers capability bundles such as wide research, website building/hosting, mobile development, and a visual Design View.

Important caution: A consistent thread across community discussions and consumer reviews is a gap between the product's autonomous execution claims and observed task completion—users report instances where the agent reported progress or completion without verifiable changes, particularly for web/app tasks.

Target market and positioning

Primary target: Individual power users, indie builders, marketers, founders, and small teams seeking agentic assistants that execute multi-step workflows (research and build).
Secondary target: SMB teams needing shared credits and concurrent task execution (Team plan).
Positioning: A general-purpose AI agent designed to execute workflows in a sandbox/virtual computer environment rather than solely answering questions.

Get a comprehensive analysis of Manus AI including market position, competitive landscape, adoption trends, and peer benchmarks.

Download full report →