Peru · AI for business · Owners, managers and teams
AI course for business in Peru
Most businesses in Peru are small, busy and sceptical of anything that sounds like a technology project. That is a healthy starting point for AI. This page is for owners, managers and team members who want to know where AI can genuinely save time in a Peruvian business, how the most useful kind of AI assistant actually understands a customer's question, what Peru's 2025 AI regulation says about telling customers, and how to train a team without stopping the business to do it.
Live classes since 2020 · 10,000+ students worldwide · every session live, online and in English
The answer in short
Modern Age Coders runs an AI course for business in Peru: live online training, taught in English, for owners, managers and staff of businesses of any size. Participants learn where AI saves time first (answering customer questions, organising documents, reading sales data), how AI tools understand meaning rather than just matching words, and how to tell customers clearly when AI is involved. No coding is needed to start. Mentors in India train one person or a whole team, one to one or in a group of fifteen to twenty, at hours that suit business opening times in Peru.
Three starting points
Which AI course suits your business?
Start with the problem you have, not the tool you have heard of. These three courses cover what most businesses need first.
About 3.38 million active private businesses in mid-2026. INEI's business register counted 3,377,839 active businesses on 30 June 2026, 1.3% more than a year earlier.
Business demography in Peru, second quarter of 2026
Measure (INEI)
Value
Note
Active businesses on 30 June 2026
3,377,839
Up 1.3% on a year earlier
Businesses created or reactivated, April to June 2026
89,846
Up 1.3% on the same quarter of 2025
Businesses closed or deregistered, April to June 2026
62,835
Up 8.1%
Birth rate
2.7%
New and reactivated businesses as a share of the stock
Closure rate
1.9%
Closures as a share of the stock
New or reactivated in Lima Metropolitana
32,890
36.6% of the national total for the quarter
Source: INEI, Demografía Empresarial en el Perú, II Trimestre de 2026, from the Directorio Central de Empresas y Establecimientos; private businesses only. Read on 9 November 2026. By our own subtraction, 89,846 minus 62,835 gives 27,011, which matches the net change INEI prints for the quarter.
What that means for AI training
Tens of thousands of businesses open in Peru every quarter, and most begin with one or two people doing everything. For them, AI is not a strategy; it is a way to answer customers at night, write a product description in English, or understand last month's sales without hiring an analyst.
A course for that audience has to be practical from the first class, and honest about what not to automate.
And for larger teams
Growing companies have a different problem: staff are already using AI tools on their own, inconsistently, sometimes with client data they should not share. Training a team together sets shared habits: what may be pasted into a tool, how output is checked, and when a customer is told.
Our Spanish page curso de IA para empresas looks at how to measure an AI pilot; this page focuses on where to start and on customers. Both describe the same live classes in English.
Where to start
Where does AI help a small business first?
Usually with words: answering repeated customer questions, drafting messages and descriptions, and finding things in the business's own documents. These are low-risk, easy to check and quick to show results.
Customer questions
Prices, opening hours, delivery areas, how to order. An assistant that finds the right answer from your own FAQ, and hands anything unusual to a person, saves hours a week. The lesson below shows how it understands the question.
Writing and translating
Product descriptions, replies, social posts, an English version of your menu or catalogue. AI drafts; a person who knows the business checks every word before it goes out.
Reading your numbers
Which products sell on which days, which customers have stopped ordering. AI help in a spreadsheet turns a sales file into a clear chart, if someone checks that the formulas are right.
What we suggest leaving until later: anything that decides about people. Peru's 2025 AI regulation, DS 115-2025-PCM, lists as high risk the use of AI to determine the selection, evaluation, hiring and dismissal of workers or applicants (article 24.1 e), and to determine people's credit assessment, except where AI is used to detect financial fraud (article 24.1 g). We cite these as published context, not legal advice; the practical message is that hiring and credit are not where a small business should experiment first.
A real lesson
How does an AI assistant understand a customer's question? Embeddings
This lesson comes from our business course. A bakery's FAQ has three answers. A customer writes "how much is a birthday cake". Matching words picks the wrong answer. Matching meaning picks the right one, and the idea behind it, embeddings, is what most modern AI assistants use.
Google's Machine Learning Crash Course describes embeddings as "low-dimensional representations of high-dimensional data, often used to capture semantic relationships between items", and adds that they "place similar items closer together in the embedding space". In plain words: each word or sentence becomes a list of numbers, and sentences that mean similar things get similar lists. Comparing two lists with a measure called cosine similarity tells you how close their meanings are, from 0 (unrelated) to 1 (pointing the same way).
Matching a customer question: words versus meaningPython
import math
# A small bakery's FAQ and one customer question, all invented by us.
faq = {
"price": "what is the cost of a whole tart or a single slice",
"hours": "what time do you open and close on sunday",
"delivery": "can you deliver a cake to miraflores and how long does delivery take",
}
question = "how much is a birthday cake"
def cosine(a, b):
dot = sum(a[k] * b.get(k, 0) for k in a)
na = math.sqrt(sum(v * v for v in a.values()))
nb = math.sqrt(sum(v * v for v in b.values()))
return dot / (na * nb) if na and nb else 0.0
def word_counts(text):
v = {}
for w in text.split():
v[w] = v.get(w, 0) + 1
return v
print("word overlap:")
for name, text in faq.items():
print(" ", name, round(cosine(word_counts(question), word_counts(text)), 2))
# A toy embedding: four hand-made dimensions (money, time, travel, food).
EMB = {"cost": (1, 0, 0, 0), "much": (0.9, 0, 0, 0), "price": (1, 0, 0, 0),
"tart": (0, 0, 0, 1), "slice": (0.2, 0, 0, 0.8), "cake": (0, 0, 0, 1),
"birthday": (0, 0.2, 0, 0.5), "time": (0, 1, 0, 0), "open": (0, 1, 0, 0),
"close": (0, 1, 0, 0), "sunday": (0, 0.8, 0, 0), "long": (0, 0.6, 0.3, 0),
"deliver": (0, 0, 1, 0), "delivery": (0, 0, 1, 0), "miraflores": (0, 0, 0.8, 0)}
def embed(text):
vecs = [EMB[w] for w in text.split() if w in EMB]
return {i: sum(v[i] for v in vecs) / len(vecs) for i in range(4)}
print("toy embedding:")
for name, text in faq.items():
print(" ", name, round(cosine(embed(question), embed(text)), 2))
The FAQ, the question and the four-dimensional embedding are invented by us for teaching. Real embeddings have hundreds of dimensions learned from data, not four chosen by hand.
OutputOutput
word overlap:
price 0.33
hours 0.0
delivery 0.34
toy embedding:
price 0.99
hours 0.11
delivery 0.28
Word overlap prefers the delivery answer, 0.34 to 0.33. The toy embedding prefers the price answer, 0.99 to 0.28.
Why word overlap fails
The customer's question and the delivery answer share the words "how", "a" and "cake". The price answer shares only "is" and "a", because it says "cost" and "tart" rather than "much" and "cake". Counting shared words, the delivery answer wins by a hair, and the customer gets told about delivery times instead of prices.
This is how simple keyword search behaves, and why so many website chat boxes give irritating answers. They match spelling, not meaning.
Why the embedding works
In the toy embedding, "much" and "cost" both point strongly along the money dimension, and "cake" and "tart" both point along the food dimension. So the question and the price answer point in almost the same direction: 0.99. The delivery answer mixes travel and time, and scores only 0.28.
Real systems learn their embeddings from huge amounts of text, so "how much" lands near "price" and "cost" without anyone writing a table. The principle is the one you just watched in twenty lines of Python.
Break it. Ask "do you open late on sunday". Which answer does each method pick? Then try a question with no words from the embedding table at all, and explain what happens.
Add a fallback. Decide what the assistant should do when even the best match is weak: hand the question to a person. Write that rule in one sentence first, then in code.
Your own FAQ. Write the five questions your customers ask most, with your answers. Those five are the start of a real assistant, and the start of an evaluation set to test it.
Spanish and English. Real multilingual embeddings place "precio" near "price". Discuss what that means for a Peruvian business serving tourists and local customers with one FAQ.
Telling customers
Do businesses in Peru have to tell customers they are using AI?
For high-risk uses, the 2025 regulation asks for transparency, and it mentions visible labelling as one way to give it. For everyday uses, telling customers is simply good practice.
Article 25.1 of DS 115-2025-PCM, published in El Peruano on 9 September 2025, asks developers and implementers of high-risk AI systems to inform users "de forma previa, clara y sencilla" about the purpose of the system, its main functions and the kind of decisions it can make. Article 25.2 says the mechanisms can include "el etiquetado visible del sistema basado en IA", a visible label telling users in advance that a product, service or content runs on AI, with an exception for AI supporting internal administrative processes without a direct impact on decisions affecting people's rights or services.
For small businesses, the regulation sets later dates for these transparency provisions: two years from publication for small companies with annual sales above 150 and up to 1,700 UIT, and three years for micro companies with annual sales up to 150 UIT. Article 18 also commits the government to promote MIPYME participation in a controlled testing environment, a sandbox, to validate AI solutions before real use. All of this is published context; what applies to a specific business is a question for its lawyer or accountant, not for us.
Say it plainly
"This assistant uses AI and may make mistakes; a person reviews orders and complaints." One sentence at the start of a chat sets the right expectation.
Offer a person
Every AI assistant should have an obvious way to reach a human. Customers forgive an AI that hands over; they do not forgive one that traps them.
Protect their data
Do not paste customers' names, numbers or orders into a public AI tool without checking its terms and your obligations. Our course covers what to anonymise and how.
Training a team
Can a team in Peru be trained in AI in English?
Yes, and for many roles the English is already there. EF's English Proficiency Index 2025 scores Peruvian adults in customer service at 585 and in marketing at 565, both in the "high" band.
EF EPI 2025, Peru, English score by job function
Job function
EF EPI 2025 score
EF band
Strategy and project management
589
High
Customer service
585
High
Marketing
565
High
Operations
554
High
Sales
547
Moderate
Administrative and clerical
540
Moderate
IT
522
Moderate
Accounting and finance
484
Low
Source: EF Education First, EF English Proficiency Index 2025, Peru country page, scores by job function, read on 9 November 2026. EF tests adults who choose to take its online test, so these scores describe test-takers in each role, not every employee in Peru.
How we organise team training
A team can learn together in one group, or in small groups by role: customer-facing staff on assistants and messages, managers on data and decisions, finance on spreadsheets and checking. Where English is weaker, as the accounting and finance score suggests it may be, one-to-one classes let the mentor slow down.
Mentors speak clearly and steadily and put key terms in the chat. We do not teach in Spanish, but the English practice is part of the value for many businesses that serve tourists or foreign clients.
Fitting it around the business
Any day, any time, flexible: choose a batch, or for a very personalised plan choose 1-on-1 at the hour that suits your family.
Lima stays five hours behind UTC all year, and India five and a half ahead, so an 08:00 class in Lima, before the shop opens, is 18:30 in India. Owners who want to code their own tools later can continue on our cursos de programación para adultos route.
Next steps
When should a business move from AI tools to AI agents?
When a task is repeated often, follows clear steps and can be checked, and when the team already uses AI tools well. Agents act in other systems, so mistakes cost more.
An assistant that answers questions is low risk: if it is wrong, a person corrects it. An agent that books, invoices or orders stock changes things, and needs limits, tests and a clear hand-off to a person. We recommend most businesses spend a few months with AI tools and assistants before building agents, and that the first agent handles something easy to undo.
When you are ready, our Spanish page agentes de IA para profesionales explains how companies keep agents under control, with an approval-gate lesson. The classes behind it are the same live classes in English. Anthropic restricts Claude to users of at least 18, so in a family business a teenage helper would follow a different route from the adults.
Progress
How a business moves from curiosity to useful AI
Our own stages for business clients. They describe the business's progress, not an individual qualification.
AI adoption stages for a business (ours)
Stage
The business can
Evidence
1. Rules
Say what may and may not be pasted into AI tools
A one-page policy staff have read
2. Drafts
Use AI for drafts that a person always checks
Ten customer messages, checked and sent
3. Data
Read its own sales or stock data with AI help
A monthly chart someone understands
4. FAQ
Answer common customer questions by meaning
The embeddings lesson, on your FAQ
5. Tell
Explain AI use to customers in one clear sentence
A label on every AI touchpoint
6. Agent
Automate one repeatable, reversible task
An agent with limits and a hand-off
For owners
Start with the question your customers ask most. That is your first AI project.
If you cannot check it, do not automate it yet.
For staff
AI writes the draft. You are still the one who knows the customer.
Matching words is not understanding. Test with a question phrased differently.
A weather forecasting site with live conditions for any location.
by Krish
Fees
What AI training costs
We charge in US dollars, to a card, per learner. Each payment is converted into soles by the card-issuing bank at its rate on the day, so your business statement may show small differences from month to month.
Reviews from learners and families on Google, unchanged.
★★★★★
"The one step solution for my son. Modern Age Coders make learning coding so simple that kids love it. The teachers explain complex concepts clearly with practical exercises and interactive content."
Ria Mukherjee
Parent
★★★★★
"Modern Age Coders has been a game-changer for me. I struggled to grasp IT concepts and coding before joining, but their classes transformed everything. I can now confidently write complex programs with ease."
Samriddha Mondal
Student
★★★★★
"One of the most wonderful education centres out there. Education is not limited to school syllabus but focuses on skill development."
Vansh Agarwal
Student
★★★★★
"My child Dhairya is really enjoying the Modern Age Coders classes. This is his first online class and he eagerly looks forward to it. I can already see his improvement, and the teachers are very cooperative."
Sonam Oswal
Parent of Dhairya
★★★★★
"Modern Age Coders have wonderful teachers who teach in a clear, easy and practical way. The teacher boosts students' confidence and inspires them to learn without hesitation."
Sonu Goyal
Parent
★★★★★
"I highly recommend this computer coding class! The teachers are incredibly knowledgeable and passionate about coding."
Ritu Kedia
Parent
Business owners ask
AI course for business: questions
Are the classes in Spanish?
No. All training is live online and in English only. Mentors speak clearly, adapt to each person's accent and pace, and put key terms in the chat. For team members whose English is weaker, one-to-one classes let the mentor slow down.
Do we need technical staff to take the course?
No. The first courses need no coding. Owners, managers and customer-facing staff learn to use and check AI tools, read their data and set up simple assistants. Technical staff can go further with Python if you wish.
Where should a small business start with AI?
With repeated customer questions, drafting messages and descriptions, and reading your own sales data. These are easy to check and quick to show results. Leave decisions about people, such as hiring or credit, until much later.
What is an embedding, in plain words?
A list of numbers that represents the meaning of a word or sentence, so that similar meanings get similar numbers. It is how an AI assistant can match "how much is a cake" to an answer about price even when no words match.
Do we have to tell customers we use AI?
Peru's DS 115-2025-PCM asks for transparency for high-risk uses and mentions visible labelling as one mechanism, with later dates for small and micro companies. That is context, not legal advice; ask your lawyer. Telling customers plainly is good practice anyway.
When are classes held?
Any day, any time, flexible: choose a batch, or for a very personalised plan choose 1-on-1 at the hour that suits your family. Peru never changes its clocks and neither does India, so the ten and a half hour gap is the same every week of the year.
Can you train our whole team?
Yes. A team can learn together or in groups by role, with exercises built on your own products and questions. Contact us with the number of people and their roles and we will suggest a plan.
Will you see our customer data?
We ask you not to share personal customer data in class. Exercises use anonymised or invented examples, and we teach your team how to anonymise data before using any AI tool.
Is there a certificate?
Each learner who completes a course receives our own certificate of completion. It is not an official Peruvian qualification, and we are not connected with INEI, PRODUCE or any Peruvian authority.
What does each person need?
Each learner needs a computer rather than a phone, reliable internet and ideally a headset with a microphone. We set up AI tool accounts together in class, choosing tools your business is happy to use.
Bring the five questions your customers ask most. In the trial, a mentor shows how an assistant would match them, where it would go wrong, and what your team would need to learn first. You leave with a recommended course and plan.
To see how we teach before booking, read how we teach or look through the courses and their syllabuses.
Training is live on video, in English, with mentors based in India. Write to us on WhatsApp in English for a quick reply.
Want a class sooner? Book a priority demo
A priority demo is a full class of about 45 to 60 minutes. After the class the mentor sends a written skill report and a personal learning roadmap on WhatsApp. If you enrol, the demo fee is adjusted against your first month's fee. Share your demo Order ID with us on WhatsApp and we take it off.
