Custom websites and AI models in Pampanga, explained.

We quote a fixed price in Philippine pesos after a free discovery call, because any number given before that conversation would be wrong for most projects. Cost is driven by the number of distinct page types, whether you need booking, payments, or user accounts, whether your team needs to edit content themselves, and how much copy and photography already exists. The proposal covers design, build, hosting setup, domain setup, and launch, priced either as a one-time figure to buy the site outright or as a monthly subscription, whichever suits how you'd rather pay.

Because a published price for "a website" is either high enough to scare off projects it would have covered, or low enough to be a bait number that rises later. Neither helps you. What we can commit to is this: the quote is fixed, not hourly, so a scope conversation never becomes a surprise invoice. And if the number doesn't work, we'll tell you what we'd cut to reach your budget rather than quietly reducing quality.

Never hourly. Whether you buy outright or subscribe, the amount is agreed before work starts with an explicit list of what's included. Hourly billing puts you and the developer on opposite sides of every decision. You hesitate to ask questions, and the incentive is to work slowly. A fixed amount means a question about scope is just a question.

Philippine pesos. Every proposal is priced in PHP by default, which keeps things simple for local clients and transparent for everyone else. If you're based abroad and would rather be billed in another currency, that's a conversation we're happy to have. Just raise it on the discovery call and we'll work out the arrangement together.

A focused marketing or portfolio site is typically two to four weeks from approved proposal to launch. Add booking systems, payments, user accounts, or third-party integrations and it runs longer. Every proposal includes a realistic timeline before work starts, and you get a live preview link you can check at any point during the build.

Yes. If you're buying the site outright, payment is normally split across milestones rather than taken in full up front, so it tracks progress you can actually see. If a large one-time figure isn't practical right now, the subscription option exists for exactly that reason: a smaller predictable monthly amount instead. Either way the schedule is set out in the proposal before anything starts.

Either. It depends on what suits you, and we work it out on the discovery call. Buying outright means one fixed price, and the site, code, and accounts are yours from launch. A subscription spreads the cost into a predictable monthly figure and bundles in hosting, maintenance, and an agreed amount of ongoing changes. Neither is the default. Businesses that want the lowest total cost and don't expect frequent changes usually buy; businesses that would rather avoid a large upfront figure, or that want someone on hand for updates, usually subscribe.

Only if you choose the subscription option. If you buy the site outright there's no ongoing fee to us. Hosting for most sites we build is free or a few dollars a month billed to you directly rather than marked up through us, and your domain renews annually at cost. If you'd rather have hosting, maintenance, and ongoing changes handled for one monthly amount, that's the subscription arrangement instead. We'll set out both figures in the proposal so you can compare them side by side.

A custom AI model is one trained or grounded on your own data rather than only on the public internet, so it can answer questions and make decisions specific to your business. You need one when off-the-shelf tools like ChatGPT can't see your internal information, get your domain wrong, or can't produce output in the exact structure your workflow requires. If a general tool already does the job, we'll tell you. That's a cheaper answer and it keeps you as a client longer.

The full range. Artificial neural networks and deep learning models for prediction, classification models, computer vision, natural language processing, time-series forecasting, and the data pipelines that feed them. Chatbots and LLM work are one part of that, not the whole of it. Two of the models in our portfolio are neural networks predicting engineering outcomes from measured inputs, with no chatbot involved anywhere.

Yes. Kleio is based in Angeles City, Pampanga, and we're glad to meet in person locally for AI projects. Most of the actual work is remote regardless of where a client is, because the collaboration that matters happens around your data rather than around a meeting table.

Yes, and sometimes that's the honest engagement. Working as an AI consultant, we'll look at where AI would actually pay for itself in your operation, which tasks are genuinely automatable, whether off-the-shelf tools already cover it, and what your data would need to look like first.

RAG (Retrieval-Augmented Generation) gives a model access to your documents at the moment it answers, so it can cite sources and stays current as those documents change. Fine-tuning further trains the model itself on your examples, so it internalises your format, tone, and domain judgement. If your goal is "the AI should know our information", you almost always want RAG, because it's cheaper and updates instantly. Fine-tuning is for when you need the model to behave a specific way.

It depends almost entirely on the state of your data. Clean, labelled records in a database make for a far smaller project than four years of scanned PDFs. We start with a free discovery call and, where useful, a short paid feasibility review on a sample of your real data, which tells you what accuracy is realistically achievable before you commit to a full build. Full projects are quoted at a fixed price in Philippine pesos.

For a retrieval system: whatever documents you already have, whether PDFs, spreadsheets, wiki pages, or email archives. For a predictive model: historical records where the outcome is known, with a few hundred rows as a workable floor depending on how many input variables there are. For computer vision: labelled images. Messy, scattered data is normal, and cleaning it is part of the project rather than a prerequisite you have to solve first.

You own the trained model weights and the surrounding code, handed over at the end of the engagement. Your training data stays yours and is never reused for another client's project. Deployment can be cloud-hosted for convenience, or run entirely on infrastructure you control when data can't leave your premises, which is often the requirement for medical, legal, financial, and government work under the Data Privacy Act.

Yes, grounded on your own documents, pricing, and policies so it answers from your business rather than guessing, with a clear handoff to a human whenever it isn't confident. That handoff is what separates a chatbot customers tolerate from one that quietly costs you enquiries.

Yes, and it's usually the highest-value version of an AI project. We build APIs and integrations connecting a model to what you already run: your website, internal tools, CRM, email, or an automation layer like n8n. AI that lives inside the workflow your team already opens every morning gets used; AI that lives in a separate app usually doesn't.

Anything rule-based and repetitive: routing and triaging enquiries, drafting standard replies, syncing records between tools, generating recurring reports, chasing invoices, extracting data from PDFs and receipts. If someone on your team does it the same way every time, it's a candidate. If it needs judgement that changes case by case, it usually isn't, and we'll say so.

Yes, and it's often the honest first step. Before anyone trains a model it's worth knowing what's actually in your data: what's predictive, what's missing, what's been recorded inconsistently. Sometimes that analysis answers the business question outright and no model is needed at all.

That's what the evaluation phase exists to answer, and it's why we recommend a short feasibility review on a sample of your real data before committing to a full build. It tells you what accuracy is realistically achievable for your specific problem, so you're deciding with a number rather than a hope. Every model we ship comes with an evaluation set and real metrics.

Yes. It's a regular part of our work and we scope it to a student budget. We handle model design, data preparation, training, evaluation, and a deployed demo for your defence, plus a walkthrough so you can explain every decision in your own words. Two of the AI projects in our portfolio were built exactly this way for graduate research.

A neural network that predicts the correct steel connection from structural design inputs, built for an engineering thesis, and a model that recommends geotextile materials from site parameters, built for a master's thesis. Both are deployed and clickable today. The common shape is prediction or classification from tabular research data, which covers most engineering, agriculture, health, and business research.

That's the part we take most seriously. We walk you through why the architecture was chosen, what each evaluation metric means, how the model compares against simpler baselines, and where it fails, because a panel will ask all four.

No. We build and explain the technical artefact (the model, the evaluation, and the deployed demo) and help you understand it well enough to write and defend it yourself. The research design, literature review, framing, and writing stay with you and your adviser.

The developer who builds your project. Kleio is a one-person studio by design, with no account manager relaying messages and no handoff between whoever scoped the work and whoever writes the code. What you explain on the call is what gets built.

A free discovery call, fifteen to forty-five minutes, with no pitch attached. You describe what you're trying to build or fix, we ask enough questions to scope it honestly, and you get one fixed-price proposal afterwards. If we're not the right fit for your project, we'll say so on the call rather than after you've paid.

If you buy the site outright, completely. The code, the design, the domain, and the content are yours with accounts in your name, and if you ever move to another developer everything transfers cleanly. On a subscription, what transfers and when is written into the agreement, and we'll walk you through it in plain language before you sign anything. Either way there's no proprietary platform holding your site hostage: it's ordinary code on ordinary hosting.

Yes, and setup for both is included with every build. We configure the hosting, connect the domain, set up SSL, and hand the accounts over in your name. What happens after that depends on how you buy. If you own the site outright, ongoing hosting is billed to you directly (free or a few dollars a month for most sites we build) and the domain renews annually at cost. On a subscription, hosting is bundled into the monthly amount and we look after it.

Every site ships with the technical SEO groundwork done: server-rendered HTML crawlers can read without executing JavaScript, unique page titles and descriptions, structured data, a valid sitemap, and fast load times. That makes your site eligible to rank. Actually ranking for competitive terms also needs a claimed Google Business Profile, consistent business listings, and content answering what people search for. We set the first up and advise on the rest.

Yes. Local ranking is mostly three things: a properly claimed and filled-in Google Business Profile, consistent business details everywhere you appear online, and a fast site with LocalBusiness structured data. The first is free and does more for local visibility than most paid advertising. We set up the technical side and walk you through the rest.

Yes, either as occasional changes when you need them or as an ongoing arrangement. Nothing is mandatory and there's no lock-in. If your site is built so your team can edit content themselves, we'll set that up and you may not need us much at all.

Yes. If you already have a brand guide, logo, or designs from another designer, we build to them faithfully. If you don't, design is included as part of the project and we'll create the visual system from scratch.

Angeles City, Pampanga, in Central Luzon, Philippines. We meet clients in person around Angeles City, the City of San Fernando, Clark Freeport Zone, and Mabalacat, and work remotely with everyone else.

Yes: Angeles City, San Fernando, Clark, Mabalacat, Mexico, Porac, Guagua, Apalit, Arayat, and Lubao. Being in the same province means a discovery call can happen over coffee rather than across a forty-message thread, which genuinely helps at the scoping stage.

Yes, and most projects run over video calls regardless. We work with clients across Central Luzon (Bulacan, Tarlac, Nueva Ecija, Zambales, Bataan) plus Metro Manila and the rest of the Philippines. Location has never limited a project.