Cisco Rolls Out AI Agents to All 90,000 Employees: A Company-Wide AI Adoption Guide for Japanese Companies in the Philippines

Cisco is distributing AI agents to all 90,000 of its employees. Drawing on this case, we take a practical look at how Japanese companies with Philippine operations can run a company-wide AI rollout — and the local pitfalls they need to watch for.

Part 1: Why This Matters

Step 1: The Philippine Business Context (3 min)

Cisco, the networking giant, has announced a plan to give AI agents to all of its roughly 90,000 employees. An AI agent does more than answer questions: it can carry out tasks on your behalf and automatically hand work off to whichever AI is best suited to the job. What makes this noteworthy is that it is not just another internal chatbot deployment — it is an effort to change how work itself gets done.

This news hits close to home for Japanese companies doing business in the Philippines. The country handles call center operations, accounting shared services, and IT help desks at massive scale, and many Japanese companies have set up local operations there. These kinds of work pair naturally with AI agents, which makes Cisco's approach of rolling out to every employee at once a useful reference for boosting the efficiency of a Philippine site.

At the same time, the Philippines has circumstances that differ from Japan: its own data privacy law, high electricity costs, and a business culture that places weight on verbal agreements. Rather than importing a foreign success story as-is, you need to adapt it to local conditions.

Picture a Japanese manager opening the morning huddle with local staff at a Manila office: "Cisco in the U.S. is apparently giving an AI assistant to all 90,000 of its people. Why don't we try it on our team too, starting with inquiry handling?" This guide lays out the concrete steps that follow that opening line.

Step 2: Key Facts from the Original Article (5 min)

We pulled out only the facts reported in the original article and organized them into a study table.

Item	Details
Scope of the rollout	Cisco is distributing AI agents to all of its roughly 90,000 employees
Start date	The rollout begins with the new fiscal year starting at the end of July
How it works	The system automatically selects the AI best suited to each task, with much of it built on the company's own on-premises infrastructure
Use in finance	AI drafts 80–90% of the first version of the MD&A narrative in financial filings
How the CFO uses it	Mark Patterson uses AI to benchmark revenue growth, EPS, R&D spending, and other metrics against peer companies
Company profile	Cisco ranks 83rd on the Fortune 500 and has been in business for more than 40 years
Stock and results	Cisco's stock is up about 53% year to date in 2026, trading at around $117 in late June
AI-related orders	The company booked $2 billion in AI-related orders in fiscal 2025 and raised its fiscal 2026 outlook to $9 billion

Source: Fortune — "Cisco is rolling out AI agents to every single one of its 90,000 employees" (July 1, 2026)

This table was compiled for learning purposes from facts in publicly available information. Please see the original article linked above for details.

Step 3: Comprehension Check (5 min)

Q1. How many employees is Cisco distributing AI agents to?

Hint: Recall the figure of roughly 90,000 that appears in the article's headline.

Q2. What is the purpose of Cisco's system that automatically picks the right AI for each task?

Hint: The article described a way to keep costs down by not relying on high-performance AI for everything.

Q3. In finance, which task does AI now draft 80–90% of?

Hint: It is the narrative section of financial filings that explains business performance in words.

Q4. What does CFO Patterson mainly use his own AI agent for?

Hint: It involves comparing his company's numbers against other companies.

Q5. What is Cisco's raised outlook for AI-related orders in fiscal 2026?

Hint: It is a big jump from the $2 billion booked in fiscal 2025.

Related: How AI Agents Help Philippine Businesses Automate Internal Operations explains this in detail.

Part 2: Putting It into Practice

Step 4: Rollout Steps in the Philippines (10 min)

Here is a step-by-step plan for running a Cisco-style company-wide AI rollout at a Philippine site without overreaching.

Step	What to do	Philippines-specific notes
1. Narrow down the target work	Start with tasks where results are easy to see, such as inquiry handling or drafting in accounting	These pair well with the country's core businesses — call centers and accounting support — making results easy to demonstrate
2. Estimate the costs	AI agents consume far more compute than ordinary chat, so model the monthly cost	Budget in pesos, and factor in the country's high electricity rates when totaling costs
3. Decide how data is handled	Draw internal lines in advance on which data may be passed to AI	Operations must comply with the rules of the NPC (National Privacy Commission), which administers the data privacy law
4. Train local staff	Hold how-to sessions and share success stories internally	Preparing both English materials and worked examples matched to local workflows speeds up adoption
5. Pilot small, then expand	Confirm the impact in one department before extending to others	Involving local managers and showing results in numbers makes internal buy-in easier

Step 5: Common Failures and How to Avoid Them (5 min)

Failure pattern 1: Forcing headquarters' policy onto the local site as-is

This is when AI usage rules decided at the Japanese head office are brought straight into the Manila site without checking local conditions. Because workflows and working languages differ, the field can't make use of it and the tools end up abandoned.

Bad: Hand out Japanese-language manuals only and close with "proceed the same way headquarters does."

Good: Work with the local IT lead to build a version adapted to Philippine workflows. Run the briefing in English with concrete examples, and always leave time for questions at the end.

Failure pattern 2: Blowing the budget by not estimating compute costs

AI agents consume far more compute than a single question-and-answer exchange. Underestimate this when budgeting and your costs will blow well past projections mid-month.

Bad: Assume "it probably costs about the same as chat" and roll out company-wide without running the numbers.

Good: Pilot in one department for a set period first and record the actual costs in pesos. Use those figures to estimate the monthly cost of a company-wide rollout.

Failure pattern 3: Launching without deciding how personal data is handled

Starting a rollout without deciding whether customer or employee data may be passed to AI risks violating the Philippine data privacy law. Fixing operations after the fact is a major undertaking.

Bad: Leave it vague which data will be used and push ahead on the field's judgment.

Good: Before rollout, document which data may and may not be shared with AI. Confirm compliance with the rules of the NPC (National Privacy Commission) before you begin.

Part 3: Going Deeper

An AI agent is an AI that does more than answer questions — it plans out the steps of a task and carries them out on your behalf. In a Philippine call center, one practical setup is to have an AI agent respond to routine inquiries automatically and escalate only the complex cases to human operators.

A token is the small unit an AI counts when reading and writing text; the more tokens used, the higher the cost. When using AI at a Manila site, estimating your monthly token usage makes it much easier to build a peso-denominated budget.

On-premises means running AI on your own equipment rather than entrusting it to an outside service. For finance-related work in the Philippines where customer data must not leave the company, going on-premises lets you operate while keeping the data in hand.

A frontier model is the most capable, cutting-edge AI available at any given time — and correspondingly more expensive to run. At a Philippine site, you can hold costs down by using inexpensive AI for simple tasks and reserving frontier models for the hard judgment calls.

A hyperscaler is one of the major cloud providers operating enormous data centers on a global scale. Companies using AI seriously in the Philippines can get started without building large local infrastructure by using these providers' services.

Step 7: Applying This to Your Company (10 min)

Pick one task to try first

Cisco is rolling out to all employees at once, but for the local site of a Japanese company, the realistic approach is to pick one task where results will be easy to see.

Something to consider: At your Philippine site, which tasks — inquiry handling, accounting drafts, and the like — are highly repetitive and easy to measure?

Next action: Ask each local department to name the one repetitive task eating the most time right now, and compile the candidates into a list.

Make AI costs visible

Because AI agents consume heavy compute, cost management makes or breaks a rollout. Build the habit of projecting costs before you use it.

Something to consider: How would you estimate total monthly costs, factoring in local electricity rates and a peso-denominated budget?

Next action: Pilot in one department for a set period, record actual costs, and calculate the monthly cost of a company-wide rollout.

Decide the data boundaries first

Starting without deciding which data may be passed to AI risks violating the Philippine data privacy law. Build internal consensus before going live.

Something to consider: How will you distinguish, among customer and employee data, what may be shared with AI and what may not?

Next action: Put the list of shareable data into a single document and have both the local manager and the headquarters contact confirm it.

Part 4: FAQ

Q1. Should we roll out to all employees at once, or start with a subset?

Cisco is distributing to 90,000 people at once, but that is an approach only possible for a large enterprise with its own infrastructure and internal training in place. For most Japanese companies with a Philippine site, it is safer to pilot in one department, confirm results, and then expand. Move in stages while watching how comfortable local staff become.

Q2. How much does it cost to use AI in the Philippines?

Exact figures depend on usage, but note that AI agents consume far more compute than ordinary chat. Electricity is expensive in the Philippines, so running on your own equipment adds that cost on top. We recommend starting small, recording actual usage in pesos, and estimating from there.

Q3. How does handling personal data differ from Japan?

The Philippines has its own data privacy law, administered by the NPC (National Privacy Commission). Its notification procedures and penalties differ from Japan's Act on the Protection of Personal Information, so proceeding on Japanese standards alone can put you out of compliance. Before rollout, decide which data goes to AI in line with local rules.

Q4. How should we train local staff?

Cisco is combining company-wide reskilling with internal sharing of use cases. In the Philippines too, holding how-to briefings and sharing successful examples works well. Prepare English materials and show concrete examples matched to local workflows, and adoption will take hold faster.

Q5. Is a verbal "we're adopting it" agreement enough?

Verbal agreements carry weight in some Philippine business settings, but for AI adoption, written agreements are essential. Especially on data handling and cost sharing, put the scope of shareable data and the allocation of responsibility on paper to prevent misunderstandings later. Make it a habit to back every verbal agreement with a document.

3 Practical Tips

Start small with one repetition-heavy task. Rather than going company-wide at once, picking a single measurable task — inquiry handling, accounting drafts — lets you verify both impact and cost quickly. Showing a small win internally makes buy-in for the next phase easier.

Make AI costs visible up front. AI agents consume heavy compute, and expanding without a trial estimate tends to blow budgets. Piloting in one department for a set period and recording actual usage in pesos lets you project the monthly cost of a company-wide rollout accurately.

Document the data boundaries before rollout. Deciding in advance which data may be passed to AI keeps operations aligned with the Philippine data privacy law. Make it standard procedure to compile the list of shareable data into one document, confirmed by both the local manager and headquarters.

Bonus: How PH AI Works Can Help

PH AI Works supports companies advancing their use of AI and technology in the Philippines. We can help you plan a company-wide AI agent rollout — the theme of this article — adapted to local Philippine conditions.

As a next step, here is what you can consult us on:

Working through which task to trial AI agents on first, based on your local operations.
Guidance on estimating costs, factoring in Philippine electricity rates and a peso-denominated budget.
Support in drawing up data handling boundaries that comply with the rules of the NPC (National Privacy Commission).

Feel free to get in touch. Initial consultations are free.

Cisco Rolls Out AI Agents to All 90,000 Employees: A Company-Wide AI Adoption Guide for Japanese Companies in the Philippines