Manus AI, a Singapore start-up, boldly proclaims the world’s first fully autonomous agent.
By Kevin Shepherdson and Harish Pillay
The Surge of Manus AI and the Frenzy for Access
Early in March 2025, the tech world was captivated by Manus AI, an artificial intelligence agent developed by a Singapore company, Butterfly Effect Pte. Ltd. Manus.AI claims to be the world’s first fully autonomous AI system. Almost immediately after its beta launch, access to Manus became an exclusive privilege, limited to an invitation-only system due to server constraints.
Butterfly Effect Pte. Ltd. is also behind another artificial intelligence application, Monica.IM.
What happened next was predictable yet astonishing. People began selling access codes on secondary markets, with reports indicating that some were priced at 999 yuan (approximately $137) to as high as 50,000 yuan ($6,900). The sheer demand for Manus is reminiscent of the early days of blockchain and cryptocurrency hype—everyone wants a piece of the next big thing, whether or not they understand how it works.
This level of frenzy raises important questions:
1. Is Manus truly revolutionary, or just another AI system riding the hype wave?
2. Can businesses trust AI agents to act autonomously without human oversight?
3. Are we witnessing the rise of truly independent AI, or are we being sold another overpromised technology?
This is why I am writing this commentary. Amidst the excitement surrounding Manus AI, the reality of AI agents needs to be critically examined. While some companies are rushing to deploy agentic AI, others are struggling to even define how AI fits into their business processes. Many organisations, especially SMEs, are still at an early stage of AI adoption—long before they should be considering fully autonomous AI.
The key question isn't whether AI agents will become mainstream, but rather: How should businesses adopt them responsibly? This requires understanding the Generalist vs. Specialist AI Agent debate and recognising that most organisations are nowhere near ready for full AI autonomy.
The Reality Check: Generalist vs. Specalist AI Agents
Manus presents itself as the future of AI-driven decision-making, showcasing an AI system (among a few others) that can analyse resumes, evaluate job candidates, make hiring decisions, and optimise business processes—all supposedly faster than a human. However, this raises critical concerns:
1. Can AI really screen resumes in seconds? What criteria does it use?
2. Does it understand industry-specific hiring needs? Or is it just keyword-matching at high speed?
3. Who ensures fairness and accuracy?
Manus represents the Generalist AI Agent approach, similar to OpenAI’s Operator and Google’s Gemini—AI systems designed to handle any task across industries. However, this only reinforces fears of job displacement, as such agents automate decision-making that has traditionally required human expertise.
By contrast, a Specialist AI approach focuses on augmenting human decision-making rather than replacing it. Instead of a one-size-fits-all solution, specialist AI solutions work within industry constraints, regulatory requirements, and human oversight to ensure ethical and practical AI deployment.
Recent developments reinforce these concerns:
1. OpenAI recently announced plans to charge businesses between $2,000 to $20,000 per AI "employee" or assistant.
2. While these specialised AI agents could help businesses, they also heighten concerns over AI replacing jobs rather than complementing them.
The Three Types of AI Applications and Their Implications
The rise of Generative AI applications has led to three broad categories of AI apps:
1. Core Apps – AI pioneers with their own foundational models and APIs, such as OpenAI, DeepSeek, and Claude.
2. Clone Apps – Start-ups that build AI solutions using APIs (Application Programming interface) from Core Apps (e.g., Manus), offering custom implementations but without original AI innovation.
3. Combination Apps – Existing enterprise software that integrates AI into their platforms (e.g., Salesforce, Microsoft, SAP).
Startups and individual developers are building on Core AI APIs rather than innovating their own models, to bring innovative applications to the market. This means that more Agentic AI solutions that look like Manus are just fine-tuned wrappers on OpenAI, DeepSeek, Claude or other LLMs.
With profit motives driving many AI start-ups, there is concern that AI systems will be rushed to market without adequate security, governance, or privacy considerations.
It would be imperative that these AI start-ups conduct rigorous testing using tools such as AI Verify Foundation’s AI governance testing framework, AI Verify, or their LLM benchmarking and evaluation toolkit, Project Moonshot, and publish their results.
Key Considerations Before Implementing AI Agents: The 6Ps Framework
Before businesses rush to deploy AI agents, there are six critical considerations:
1. Purpose – What is the AI agent designed to do? Avoid scope creep where AI exceeds its intended role.
2. Permissions – What level of access does the AI have? AI should not act without explicit user or organisational approval.
3. Passwords – AI agents need secure authentication to prevent unauthorised access to systems.
4. Payments – AI agents handling transactions must comply with regulatory and fraud prevention standards (e.g., PSD2, AML).
5. Privacy – AI must comply with data protection laws such as GDPR, PDPA, CCPA to prevent misuse.
6. Protection – AI must be safeguarded against data breaches, adversarial attacks, and misinformation risks.
Without governance in place, AI autonomy could cause more harm than good.
A Roadmap for Responsible AI Adoption: The AI Maturity Model
Many organisations are still grappling with AI’s potential use cases, let alone thinking about full automation. Businesses need a structured roadmap before transitioning from basic AI applications to fully autonomous AI agents.
A Gen AI Maturity Model can help companies progress through the following stages:
Stage | Gen AI Adoption | AI Agent Adoption |
1. Use | Exploring Gen AI benefits | Companies use basic AI assistants for knowledge retrieval and automation. |
2. Create | Building chatbots & AI tools | Developing workflow-driven AI Assistants and that pass conversation history between apps. |
3. Deploy | Deploying AI to users & create value | Deploy AI Assistant and structuring AI Agents to execute business processes. |
4. Govern | Ensuring AI governance & compliance | Defining decision-making prompts, execution rules, and audit trails for AI-driven workflows. |
5. Manage | Managing AI transformation end-to-end | Implementing autonomous agents with clear business controls. |
Most companies today are still between Stage 1 (Use) and Stage 2 (Create)—far from ready to deploy fully autonomous AI.
The Straits Interactive Approach: AI as a Thinking Tool, Not a Replacement
At Straits Interactive, we reject the idea that AI should replace human expertise. Instead, we focus on helping organisations augment human intelligence with AI, ensuring:
1. AI enhances decision-making, rather than taking over.
2. AI aligns with business strategy, regulatory requirements, and ethics.
3. Companies develop AI workflows that fit their industry’s unique challenges.
Our structured generative AI education adoption process (especially in our Generative AI, Ethics and Data Protection courses and the recent Advance Certificate in Generative AI Apps Design and Prompt Engineering) includes:
1. Prompt Engineering – Using AI to extract insights while retaining human control.
2. Workflow Agents – Passing conversation history across apps for context-aware automation.
3. AI Governance – Defining rules, compliance measures, and oversight mechanisms.
4. Embedding AI into Business Workflows – Aligning AI with business processes, not just generic automation.
5. Scaling Towards Responsible AI Autonomy – Ensuring AI acts within ethical and regulatory boundaries.
Unlike Manus’s "snap-your-fingers" AI automation, our approach ensures AI adoption is practical, sustainable, and trustworthy.
The Road Ahead For AI Agents
The hype surrounding Generalist AI Agents may dominate the headlines, but businesses need to approach AI adoption with caution and strategic intent.
The future of AI is not about full autonomy, but about responsible integration—where AI augments human intelligence rather than replaces it.
The key to success? Understanding AI’s role, ensuring governance, and adopting AI in a way that aligns with business needs, ethics, and regulations.