When AI Agents Start Writing the Policies, What Is Left for People to Do? | Case Study for Japanese Companies in the Philippines

Internal policies, procedure manuals, compliance frameworks — there is a category of work everyone agrees matters and everyone keeps postponing. AI agents are now moving into it in the enterprise market. In this module, we use a partnership between a compliance software company and an AI implementation firm to think about how much of policy and procedure work can be handed to AI, and what people must keep hold of to the end, applied to the situation of Japanese companies in the Philippines.

According to reports, on 27 July 2026 LogicGate, which provides compliance software, announced a partnership with Ode, an enterprise AI implementation company that uses Anthropic's models. The stated aim is to accelerate how LogicGate's customers build, actually operate, and update their GRC (governance, risk and compliance) programs. Ode is reported to have been launched in 2026 through a partnership involving Anthropic, Blackstone, Hellman & Friedman and a consortium of investors. In other words, AI agents have begun moving in earnest into the world of documents and procedures. In this module, we work through what that means in Parts 1-4, applying it to your own situation.

---

Part 1: Read to Consider the Implications for Your Company

Step 1: Pre-Reading (3 min)

Before reading, put yourself in your own company's shoes.

• When were your internal policies and procedure manuals last updated?
• How many policies do you have that were written and never read?
• Is the reason policy work gets postponed a lack of time, or not knowing how to write them?

Step 2: First Reading (10 min)

Below is a fictional internal memo, written from the perspective of a Japanese company in the Philippines, based on the facts of the news.

---

Internal Memo: Can We Use AI for Policy and Procedure Work?

LogicGate, a compliance software company, has announced a partnership with Ode, which uses Anthropic's models. The aim is reported to be speeding up how customer companies build, operate, and update their GRC programs.

The first thing worth noting is that the scope covers not only building but operating and updating. Where we stumble on policy work is not the initial drafting so much as the stage afterwards, when nothing gets updated and the document quietly goes out of date. If that stage is being addressed, this is a meaningful change.

Second, this is a world of documents and procedures. Policies, checklists, risk registers — all of them are made of text and structure. That is a shape AI handles comparatively well, and the same thinking is available to us.

Third, an enterprise-scale mechanism does not simply arrive at our door. What is reported is a large enterprise initiative; at our size this is not a conversation about buying the same product. But the approach is borrowable: AI drafts, people decide, and the trigger for updating is agreed in advance.

Three implications for us: (1) list the policies we have been postponing and have AI draft them; (2) require that judgement and approval always sit with a person; (3) decide the trigger for updates in advance so documents do not go stale.

---

Source: Ode with Anthropic and LogicGate Announce Partnership to Scale Expert-Led Client Outcomes through Accelerated Agentic Capabilities (27 July 2026)

Note: The business scenario above is a fictional internal memo created for learning purposes from publicly available reports. The details of the partnership may change; please check the primary source linked above for the latest information.

Step 3: Comprehension Check (5 min)

• What does the partnership say it will do with GRC programs?
• What kind of company is Ode reported to be?
• Which domain does this development show AI agents entering?

Step 4: 3-Minute Briefing (10 min)

Practise explaining this to your management meeting in three minutes. It lands best in this order: what happened (AI agents entering the policy and risk-management domain), why it matters (it is exactly the area we keep postponing), and what we should do (AI drafts, people decide, agree the update trigger).

Related: What a Major IT Services Firm's AI Partnership Reveals: The Three Things That Close the Gap Between a Capable Model and an Actual Result | Case Study for Japanese Companies in the Philippines explains this in detail.

Part 2: Key Terms Explained (for Executives)

GRC (governance, risk and compliance) — A collective term for how a company is governed, how it manages risk, and how it keeps to laws and rules. In plain terms: the system for managing your internal rules and whether they are actually followed.

Operating a policy — The state in which a written policy is actually used. A document existing and a document being operated are different things, and most companies stumble on the second.

AI agent — An AI that takes an instruction and works through several steps on its own. The difference from earlier use is that it does not just write one passage; it gathers material, organises it, and produces a draft end to end.

Update trigger — An agreed signal for reviewing a policy or procedure. Deciding on something like "when the law changes" or "when the same problem happens twice" makes it much less likely that stale policies are left in place.

Related: Intelligence Is No Longer the Constraint: What the Change in the Industry's Watchword Means | Case Study for Japanese Companies in the Philippines explains this in detail.

Part 3: Applying It to Your Company

List the documents you have been postponing

Start by writing down the documents you know you should produce but have not. Employment rules, information-handling rules, procedure manuals, agreements with subcontractors — you will probably list several. Before any conversation about handing work to AI, be clear about the target.

Related: Lessons from a Government Order to Stop Using an AI Provider: Preparing for the Day Your AI Becomes Unavailable | A Case Study for Japanese Companies in the Philippines explains this in detail.

Decide that AI drafts and people judge

Policy text is well suited to AI drafting. What must stay with people is whether the content is sound, whether it matches your actual operation, and any judgement touching the law. If you operate in the Philippines, this needs checking against local employment practice and relevant regulation, which is territory for a professional adviser. Put the division — AI produces the shape quickly, people guarantee the correctness — in writing at the outset.

Agree the update trigger first

Policies go stale because nobody is responsible for updating them. Set an annual review month, check when you see reports of a legal change, or fix the relevant section when the same problem occurs twice — any one of these prevents neglect. Because AI makes the updating work itself lighter, this habit is more realistic than it used to be.

Part 4: Common Failure Patterns (What Not to Do)

Failure 1: Adopting an AI-written policy as it stands

Because it looks well organised, it gets adopted without the content being checked. AI is good at producing conventional structure; it cannot judge whether the result fits your operation or local law. Note that the tidier the document, the more tempting it is to skip the check.

Failure 2: Translating the Japanese policy without reflecting local conditions

Translating head office's policy directly produces sections that do not match local employment practice or procedure. A translation is a starting point, not a finished product. Revision against local practice is always required.

Failure 3: Treating the document as the finish line

A policy existing and employees knowing about it are different things. If you have not decided where it will live and when it will be communicated, the work itself is wasted. Decide where it goes and when you will tell people before you start writing.

Failure 4: Trying to sort everything at once

Widen the scope too far and you run out of energy partway. Pick one area — high volume, or somewhere you have actually been burned. Completing one builds internal understanding and makes the next one easier.

Practical Tips (3 Tips)

1. Start where you have actually been burned. Policies covering an area where a real problem occurred win internal agreement more easily and their value is felt. Start from the pain rather than the ideal.
2. Always give the AI your own circumstances. Headcount, sites, industry, local practice — draft without supplying context and you get a generic document you cannot use. Time spent explaining the premises improves the result.
3. Decide who reviews before you begin. Produce a draft without deciding who reads and approves it and it will sit unreviewed. Name the reviewer before you start.

Bonus: How to Use the PH AI Works Free Consultation

PH AI Works supports Japanese companies in the Philippines, in Japanese, from designing how to use AI for internal policies and procedures to building the review structure and embedding it in existing workflows. If you have documents you know you should produce but have never started, feel free to use our free consultation. We can begin together with a quick assessment of your current setup.

Sources

• Ode with Anthropic and LogicGate Announce Partnership to Scale Expert-Led Client Outcomes through Accelerated Agentic Capabilities (27 July 2026)