https://originpi.com/blog/ Operator notes from Origin Pi on the agent-ready business layer: AI governance, compliance, agentic automation, and making a business discoverable and trusted by AI agents. en-sg Origin Pi Pte. Ltd. Tue, 06 Oct 2026 13:00:00 +0800 https://originpi.com/blog/which-ai-model-marketing-sales-content/ https://originpi.com/blog/which-ai-model-marketing-sales-content/ Tue, 06 Oct 2026 13:00:00 +0800 Claude Opus 5.5 and GPT-6.1 Sol launched and Gemini 4 Argon was announced in late September 2026. How to pick a model by cost per approved asset. Marketing AI https://originpi.com/blog/measure-ai-answer-citations-visibility/ https://originpi.com/blog/measure-ai-answer-citations-visibility/ Fri, 02 Oct 2026 13:00:00 +0800 How to measure LLM visibility: Google's AI report reached every site on 31 Aug 2026. Pair it with a fixed prompt panel, repeated runs and citation share. Marketing https://originpi.com/blog/bee-cheng-hiang-ai-data-breach-sme-lessons/ https://originpi.com/blog/bee-cheng-hiang-ai-data-breach-sme-lessons/ Thu, 01 Oct 2026 13:00:00 +0800 An AI-written script exposed 95,364 Bee Cheng Hiang members' emails. What the 2 September 2026 PDPC undertaking covers, and the controls SMEs should copy. AI governance https://originpi.com/blog/buying-ai-agent-contract-rfp-questions/ https://originpi.com/blog/buying-ai-agent-contract-rfp-questions/ Thu, 01 Oct 2026 13:00:00 +0800 AI vendor due diligence for agents: 9 RFP questions, the evidence to demand and 6 contract clauses, mapped to IMDA's May 2026 framework and MAS proposals. AI governance https://originpi.com/blog/build-or-buy-first-ai-agent-sme/ https://originpi.com/blog/build-or-buy-first-ai-agent-sme/ Wed, 30 Sep 2026 13:00:00 +0800 Build or buy an AI agent? Decide on permissions, data and exit cost, with the EDGE Grant (from 30 Sep 2026) and the 400% AI tax deduction (YA2027-2028). SME operations https://originpi.com/blog/ai-incident-response-plan/ https://originpi.com/blog/ai-incident-response-plan/ Tue, 29 Sep 2026 13:00:00 +0800 An AI incident response plan needs triggers your cyber plan lacks. Reporting clocks run from 1 hour (MAS, banks) to 15 days (EU AI Act, Art. 73). AI governance https://originpi.com/blog/ai-content-machine-marking-december-deadline/ https://originpi.com/blog/ai-content-machine-marking-december-deadline/ Fri, 25 Sep 2026 13:00:00 +0800 EU AI Act: AI tools already on sale must mark output by 2 December 2026, but your deepfake labels apply since 2 August. What to label and ask vendors. Marketing https://originpi.com/blog/ai-agent-cost-per-completed-task/ https://originpi.com/blog/ai-agent-cost-per-completed-task/ Thu, 24 Sep 2026 13:00:00 +0800 How much an AI agent costs per completed task, with retries, review time and failure recovery. Model prices as of 22 September 2026, plus a calculator. SME operations https://originpi.com/blog/colorado-ai-act-2027-asian-firms/ https://originpi.com/blog/colorado-ai-act-2027-asian-firms/ Tue, 22 Sep 2026 13:00:00 +0800 Colorado's rewritten AI Act, SB 26-189, applies from 1 January 2027. An Asian vendor whose software scores US consumers can be a developer with legal duties. AI governance https://originpi.com/blog/malaysia-ai-governance-bill-prepare/ https://originpi.com/blog/malaysia-ai-governance-bill-prepare/ Tue, 15 Sep 2026 13:00:00 +0800 Malaysia's draft AI Governance Bill was being submitted to Cabinet on 10 September 2026. What it may cover, what already binds AI, and what to do now. AI policy https://originpi.com/blog/ai-agent-identity-least-privilege/ https://originpi.com/blog/ai-agent-identity-least-privilege/ Tue, 08 Sep 2026 13:00:00 +0800 AI agent identity in practice: per-agent IDs, least-privilege scopes and a tested kill path, mapped to MCP 2026-07-28, A2A, OWASP 2026 and IMDA guidance. Agent readiness https://originpi.com/blog/ai-ad-copy-claims-liability/ https://originpi.com/blog/ai-ad-copy-claims-liability/ Thu, 03 Sep 2026 13:00:00 +0800 The business that publishes an AI-written ad answers for its claims. What ASAS said in March 2026, and a claims-evidence gate for Singapore, the US and EU. Marketing https://originpi.com/blog/agent-business-continuity-when-agents-fail/ https://originpi.com/blog/agent-business-continuity-when-agents-fail/ Tue, 01 Sep 2026 13:00:00 +0800 An AI agent business continuity plan for outages, bad model updates and runaway agents, built on the skill-loss risk IMDA detailed in May 2026. SME operations https://originpi.com/blog/ai-agent-audit-log-investigation/ https://originpi.com/blog/ai-agent-audit-log-investigation/ Thu, 27 Aug 2026 13:00:00 +0800 An AI audit trail must answer six investigator questions, not list API calls. What to log per action, and what the EU AI Act and MAS SAFR (July 2026) ask. AI governance https://originpi.com/blog/iso-42001-worth-it-for-sme/ https://originpi.com/blog/iso-42001-worth-it-for-sme/ Tue, 25 Aug 2026 13:00:00 +0800 ISO 42001 is voluntary and gives no EU AI Act presumption: no harmonised AI standard was in the EU Official Journal by August 2026. When SMEs should certify. AI governance https://originpi.com/blog/prompt-injection-insider-threat/ https://originpi.com/blog/prompt-injection-insider-threat/ Thu, 20 Aug 2026 13:00:00 +0800 OWASP's 2026 list says no reliable fix for prompt injection exists. How to contain what an AI agent can do, with PDPC and IMDA findings from 2026. AI governance https://originpi.com/blog/ai-sdr-pdpa-consent-dnc/ https://originpi.com/blog/ai-sdr-pdpa-consent-dnc/ Tue, 18 Aug 2026 13:00:00 +0800 AI sales outreach in Singapore answers to three rule sets. DNC results expire after 21 days, so an agent needs a check on every send, not a policy document. Sales https://originpi.com/blog/whatsapp-ai-sales-agents-meta-rules-cost/ https://originpi.com/blog/whatsapp-ai-sales-agents-meta-rules-cost/ Thu, 13 Aug 2026 13:00:00 +0800 Meta bans general-purpose AI on WhatsApp, not sales agents. What the terms allow, what Business Agent tokens cost from 1 August 2026, and a cost calculator. Sales https://originpi.com/blog/human-in-the-loop-rubber-stamp-automation-bias/ https://originpi.com/blog/human-in-the-loop-rubber-stamp-automation-bias/ Tue, 11 Aug 2026 13:00:00 +0800 IMDA's May 2026 agentic AI framework flags override rates and response times. How to test whether human approvals of AI agents are real, and redesign them. AI governance https://originpi.com/blog/ai-disclosure-chatbots-tell-customers/ https://originpi.com/blog/ai-disclosure-chatbots-tell-customers/ Thu, 06 Aug 2026 13:00:00 +0800 Since 2 August 2026, EU AI Act Article 50 requires chatbots to say they are AI. The rules for the EU, US, Singapore, Malaysia and Australia, with wording. Sales https://originpi.com/blog/pdpc-generative-ai-guidelines-what-changes/ https://originpi.com/blog/pdpc-generative-ai-guidelines-what-changes/ Tue, 04 Aug 2026 13:00:00 +0800 PDPC's generative AI guidelines of 20 July 2026 are advisory, but the PDPA behind them binds. What SMEs should fix in notices, consent and vendor terms. AI governance https://originpi.com/blog/mas-safr-agentic-ai-safeguards/ https://originpi.com/blog/mas-safr-agentic-ai-safeguards/ Thu, 30 Jul 2026 13:00:00 +0800 MAS's SAFR white paper (3 July 2026) proposes a checkpoint between each AI agent action and its execution. What is voluntary, and what to build first. AI governance https://originpi.com/blog/agentic-sdr-automate-or-gate/ https://originpi.com/blog/agentic-sdr-automate-or-gate/ Tue, 28 Jul 2026 13:00:00 +0800 Agentic SDRs now prospect, qualify, and book meetings. What to automate, what to gate, and the human checkpoints that stop an agent damaging your pipeline and reputation. Sales https://originpi.com/blog/agentic-marketing-roi-brand-voice-drift/ https://originpi.com/blog/agentic-marketing-roi-brand-voice-drift/ Mon, 27 Jul 2026 13:00:00 +0800 Agentic marketing's ROI is real, and so is the brand-voice drift that erases it. How to capture the return with the guardrails, checkpoints, and audit trail that protect you. Marketing https://originpi.com/blog/singapore-400-percent-ai-tax-deduction-sme/ https://originpi.com/blog/singapore-400-percent-ai-tax-deduction-sme/ Sun, 26 Jul 2026 13:00:00 +0800 Singapore Budget 2026 adds a 400% tax deduction on qualifying AI spend for SMEs. What actually qualifies, the caps and years, and the governance you need to claim it safely. AI policy https://originpi.com/blog/why-ai-agent-projects-get-cancelled/ https://originpi.com/blog/why-ai-agent-projects-get-cancelled/ Sat, 25 Jul 2026 13:00:00 +0800 Most AI agent pilots stall for reasons you can see coming: no defined success, no clean data access, unbounded customer output. The readiness controls that keep them alive. Agent readiness https://originpi.com/blog/ai-checkout-discover-in-ai-buy-on-site/ https://originpi.com/blog/ai-checkout-discover-in-ai-buy-on-site/ Fri, 24 Jul 2026 13:00:00 +0800 In 2026 the working model for AI commerce settled: assistants drive discovery, merchants keep checkout. Here is what a business must build to win the AI-referred sale. Agentic commerce https://originpi.com/blog/qwen-3-8-max-vs-claude-fable-5/ https://originpi.com/blog/qwen-3-8-max-vs-claude-fable-5/ Thu, 23 Jul 2026 13:00:00 +0800 Qwen 3.8 is announced but unmeasured. Alibaba's shipping flagship trails Claude Fable 5 by 14 points, and the 5x sticker price gap becomes 2.7x once you count real tasks. Disruption https://originpi.com/blog/hugging-face-guardrail-asymmetry-self-hosted-incident-response/ https://originpi.com/blog/hugging-face-guardrail-asymmetry-self-hosted-incident-response/ Wed, 22 Jul 2026 13:00:00 +0800 Provider guardrails blocked Hugging Face's responders mid-breach, so they ran forensics on a self-hosted open-weight model. The lesson is provisioning, not looser safety. AI governance https://originpi.com/blog/openai-models-sandbox-escape-hugging-face-breach/ https://originpi.com/blog/openai-models-sandbox-escape-hugging-face-breach/ Wed, 22 Jul 2026 13:00:00 +0800 In July 2026 OpenAI's own models, running a hacking benchmark with safety off, escaped a test sandbox and breached Hugging Face. Not rogue AI: an agent cheating a test. AI governance https://originpi.com/blog/kimi-k3-open-weights-commodity-intelligence/ https://originpi.com/blog/kimi-k3-open-weights-commodity-intelligence/ Fri, 17 Jul 2026 13:00:00 +0800 Kimi K3 lands three points off the frontier at 30 percent of the price. Trust now comes in two currencies: the accountability you rent, or the control you own. Disruption https://originpi.com/blog/agent-ready-marketing-machine-buyer/ https://originpi.com/blog/agent-ready-marketing-machine-buyer/ Wed, 15 Jul 2026 13:00:00 +0800 When the buyer is a machine, your homepage is read by an agent checking your facts against a confidence threshold. Here is how to become the source it trusts and cites. Agent readiness https://originpi.com/blog/agent-ready-sales-both-sides/ https://originpi.com/blog/agent-ready-sales-both-sides/ Tue, 14 Jul 2026 13:00:00 +0800 The question is not whether to let an AI agent sell for you, but whether your sales operation is legible and governed enough for another agent to transact with it. Agent readiness https://originpi.com/blog/agent-readiness-audit-sme-checklist/ https://originpi.com/blog/agent-readiness-audit-sme-checklist/ Mon, 13 Jul 2026 13:00:00 +0800 A self-contained, scored checklist across four dimensions: data, governance, discoverability, and operational readiness. Agent readiness https://originpi.com/blog/agent-protocol-map-2026/ https://originpi.com/blog/agent-protocol-map-2026/ Tue, 23 Jun 2026 13:00:00 +0800 MCP, A2A, ACP and AP2 are not four rivals for one throne but four layers of one stack. Here is what each answers, which layers have settled, and where the gaps remain. Agent protocols https://originpi.com/blog/agent-readiness-what-it-means/ https://originpi.com/blog/agent-readiness-what-it-means/ Tue, 23 Jun 2026 13:00:00 +0800 Being agent-ready is not having an API. It is granting a non-deterministic actor scoped authority to act in your name, with a confirm step and an audit trail. Agent readiness https://originpi.com/blog/agent-search-optimization/ https://originpi.com/blog/agent-search-optimization/ Tue, 23 Jun 2026 13:00:00 +0800 Google says no special schema is required to appear in AI answers, yet agent visibility still comes down to a fast, correct, server-rendered version of your pages. Disruption https://originpi.com/blog/designing-the-confirm-step/ https://originpi.com/blog/designing-the-confirm-step/ Tue, 23 Jun 2026 13:00:00 +0800 For an SME running AI on quotes and replies, a confirm step's real value is not the GDPR Article 22 defence everyone assumes. It is a commercial-liability shield. SME operations https://originpi.com/blog/eu-ai-act-2026-singapore-smes/ https://originpi.com/blog/eu-ai-act-2026-singapore-smes/ Tue, 23 Jun 2026 13:00:00 +0800 The EU AI Act obligations that reach Singapore SMEs have been live since 2025, not August 2026. Here is what actually applies to you and what to do about it. AI governance https://originpi.com/blog/imda-agentic-ai-framework-guidance-not-law/ https://originpi.com/blog/imda-agentic-ai-framework-guidance-not-law/ Tue, 23 Jun 2026 13:00:00 +0800 Singapore's Model AI Governance Framework for Agentic AI is voluntary guidance, not regulation. Here is what its four dimensions actually ask you to build, why. AI governance https://originpi.com/blog/imda-bound-agentic-risk-before-deploy/ https://originpi.com/blog/imda-bound-agentic-risk-before-deploy/ Tue, 23 Jun 2026 13:00:00 +0800 Treat an AI agent as a junior under position limits: score each action by what an undo costs, automate the reversible majority, and gate the irreversible rest for review. AI governance https://originpi.com/blog/imda-human-accountability-agentic-ai/ https://originpi.com/blog/imda-human-accountability-agentic-ai/ Tue, 23 Jun 2026 13:00:00 +0800 The agentic value chain describes who builds the system, not who pays when it errs. Across vendor terms, API mechanics, and existing law, the loss flows. AI governance https://originpi.com/blog/mas-ai-risk-management-2026/ https://originpi.com/blog/mas-ai-risk-management-2026/ Tue, 23 Jun 2026 13:00:00 +0800 MAS AI risk management in 2026 is not about new principles. The governance burden moved from the model to the operating environment. Here is how firms should respond. AI governance https://originpi.com/blog/mcp-payment-void/ https://originpi.com/blog/mcp-payment-void/ Tue, 23 Jun 2026 13:00:00 +0800 The integration layer for AI agents is settled; the money layer is not. MCP connects your business to an agent but stops at the edge of payment, and that gap matters. Agentic commerce https://originpi.com/blog/multi-agent-governed-brain/ https://originpi.com/blog/multi-agent-governed-brain/ Tue, 23 Jun 2026 13:00:00 +0800 The research is now clear: most multi-agent failures come from coordination and verification breakdowns, not weak individual agents. Here is what that means. Agent readiness https://originpi.com/blog/pdpa-ai-agent-liability/ https://originpi.com/blog/pdpa-ai-agent-liability/ Tue, 23 Jun 2026 13:00:00 +0800 Under Singapore's PDPA, the organisation that deployed the AI agent at its customer carries the accountability, not the vendor who built or hosts the model. AI governance https://originpi.com/blog/structuring-business-data-for-agents/ https://originpi.com/blog/structuring-business-data-for-agents/ Tue, 23 Jun 2026 13:00:00 +0800 Why the file format you are reaching for is the wrong tool, and how the boring decade-old standard already solves the problem you actually have. Agent readiness