FAQ
62 questions, answered the way we would answer them on a call: what Hidook is, how a project runs, what it costs, who owns the code, and how we build AI features that hold up in production.
Hidook (hidook.tech) is a remote-first product engineering studio. It designs, builds and runs custom web applications, SaaS platforms, MVPs, AI and LLM systems and cloud infrastructure for founders and companies in Europe, the UK and the US.
Hidook takes a software product from the first sketch to production and keeps it running: discovery, architecture, design, engineering, launch and ongoing scaling, with a working release on a live URL every Wednesday.
Founders who need an MVP built properly, companies replacing spreadsheets and e-mail threads with a real application, teams turning a service into a subscription product, and businesses that want to add AI features that are measured rather than guessed.
Hidook is Hidook Ltd, a company registered in England and Wales (company number 15995875) with its registered office in Romford, UK. The team is remote-first on purpose, works with clients in Europe, the UK and the US, and always keeps at least four working hours in common with the client’s time zone.
Hidook Ltd was incorporated in England on 3 October 2024.
Senior engineers do the work (the people on the sales call are the people writing the code), clients see a live demo every Wednesday instead of status reports, clients own the repository and cloud accounts from day one, and no AI feature reaches production without an evaluation set and a measured cost per answer.
Yes. Hidook designs, builds and runs Instafidget (instafidget.com), Instagram feed widgets for any website, and Hidook Site Builder (lp.hidook.agency), a website builder for restaurants, salons, trades and local services.
Client and group projects include QualsVault (qualsvault.com), Business Summit London (rotarielena.me), Renoveasy (renoveasy.co.uk), Hidook Group (hidook.com), alongside Hidook’s own products Instafidget and Hidook Site Builder. Each case study is at https://hidook.tech/work/.
Building an app when a spreadsheet or an off-the-shelf tool would do, going six months without a demo, juniors learning on the client’s budget, code the client cannot take with them, and AI features nobody has measured.
Write to no.bs@hidook.tech or use the form at https://hidook.tech/contact/. A senior engineer, not a sales rep, replies within one working day.
Five steps. Discovery (week 0): workshops, constraints and the one metric that matters, ending in a written scope and a price. Blueprint (week 1): architecture, data model and a clickable prototype, with decisions recorded as ADRs. Build (weekly): a demo and a deploy every Wednesday. Launch: load tests, security review, monitoring and a launch-day runbook. Scale: performance work, new features or a clean handover.
As soon as the scope is agreed. Discovery usually starts the Monday after.
Short workshops to understand the users, the constraints and the single metric the product must move. The output is a written scope, a first architecture and a price, so the decision to build is made on paper before any code is written.
Monday: plan the week and cut scope out loud. Wednesday: a demo and a deploy to a live URL that the client can click. Friday: code review and a check that production is healthy before the weekend.
Because a release needs watching. Deploying mid-week leaves two working days to see how real users use it and fix anything before the weekend, instead of leaving problems to whoever is on call on Saturday.
Every change gets its own preview URL, and every Wednesday there is a demo on a live link. If there is nothing to show on a Wednesday, the client hears why on that Wednesday, not at the end of the project.
An Architecture Decision Record: a short document that records one technical decision, the options considered and why one was chosen. Hidook writes them so a client or a future team can understand the system in minutes.
Yes. Hidook regularly pairs with internal engineers, reviews their code and aims to leave the team stronger than it found it.
Hidook stays on call through launch and then either keeps improving the product as a monthly team or hands it over cleanly to the client’s own engineers, with documentation and tests in place.
Europe, the UK and the US. Hidook keeps at least four working hours in common with every client.
Three ways. Sprint: a fixed price and a fixed date for discovery, a prototype or an MVP, set after discovery. Product team: a monthly fee for a dedicated squad of engineers, a designer and a lead. Partnership: a monthly retainer for fractional CTO work, architecture reviews and senior help.
Both, depending on the stage. Discovery and MVP sprints have a fixed price because the scope is written down first; ongoing product work runs as a monthly team because priorities change as the product meets users.
It depends on scope, the number of user roles, integrations, platforms and compliance needs. Hidook’s contact form uses four bands as orientation: under €25k, €25–75k, €75–150k and €150k+. Discovery turns a rough idea into a written scope and an exact price.
Trying to validate several ideas at once, multiple user types that interact with each other, integrations with systems nobody in the room controls, native mobile apps, strict compliance, data migration and unclear decision-making. Each one adds screens, edge cases and testing.
Write the one question the MVP must answer, start with a single user type, run the admin side on a spreadsheet or existing tool at first, start on the web, and do integrations by hand for the first users when that is safe.
Yes: hosting, monitoring, third-party APIs, AI model usage and maintenance. Hidook sets error tracking and cost monitoring up before the first user so these costs are visible from day one.
Mostly TypeScript with React or Next.js on the front end, Node.js or Go on the back end and PostgreSQL for data. We pick what your team can maintain after us.
Yes. We start with a short code and architecture review, then agree on what to keep, what to refactor and what to replace.
Yes: accounts and teams, roles, subscription billing with Stripe, audit logs and an admin console are standard parts of our SaaS work.
It depends on scope, which is why discovery comes first. We agree on a fixed scope and date before the build starts, and cut scope rather than move the date.
Yes. With React Native and Expo we ship iOS, Android and web from one codebase, which keeps an MVP fast and affordable.
You choose: we keep building with you, or we hand over with documentation and help you hire your own engineers.
We work with Anthropic Claude, OpenAI and open-weight models, and choose per task based on quality, latency, cost and where your data may be processed.
Data stays in your cloud accounts where possible, access is scoped per user, and we use providers and settings that do not train on your data.
Answers are grounded in retrieved sources and cite them, uncertain answers are flagged, and a person approves anything that leaves the company.
AWS, Google Cloud and Cloudflare most often, with Terraform so the setup is documented and reproducible.
Usually, yes. We start by measuring where the money goes, then right-size, remove idle resources and move the right workloads to cheaper services.
Yes, for products we build or take over, with agreed response times and runbooks your own team can follow.
We can, but we are best at design that ships. When we only design, we still deliver components and specs your developers can use directly.
Yes. Prototypes go in front of real users before they reach engineering, so we change pixels instead of code.
Yes. We audit against WCAG and fix the issues that matter most for your users first.
Sets the technical direction, reviews architecture and code, helps hire and onboard engineers, and represents technology to investors and partners — a few days a month.
Yes. You get a written, plain-language review of code quality, security and risks, with a prioritised list of what to fix.
Yes: role definitions, technical interviews and onboarding, so new engineers are productive in weeks, not months.
Service pages: Web app and SaaS development · MVP development for startups · AI and LLM integration · Cloud and DevOps services · Product design for web and mobile apps · Fractional CTO and technical partnership
The client, from the first commit. Everything lives in the client’s own repository and cloud accounts.
Yes. Documentation is handed over, not promised, tests run on every pull request and the infrastructure is defined as code, so another team can take over without starting again. Hidook’s rule is that a client should be able to fire Hidook and keep going.
Yes. Hidook starts with an access and ownership audit and a short code and architecture review, then agrees with the client what to keep, what to refactor and what to replace.
The client owns the repository from day one, tests run on every pull request, every change has a preview URL, error tracking is in place before the first user, and documentation is delivered with the code.
Mostly TypeScript, React and Next.js (or Astro for content sites) on the front end; Node.js, Go or Python on the back end; PostgreSQL and Redis for data; AWS, Google Cloud or Cloudflare with Terraform for infrastructure; GitHub Actions for CI/CD; Grafana for observability. The choice is always what the client’s team can maintain afterwards.
Yes. Hidook uses React Native with Expo to ship web, iOS and Android from one codebase, and recommends starting with the web or an installable web app when users mostly sit at a desk.
Yes: accounts and teams, roles and permissions, subscription billing with Stripe, audit logs and an admin console are standard parts of Hidook’s SaaS work.
Yes: infrastructure as code with Terraform, CI/CD pipelines, preview deploys for every pull request, monitoring, alerting and on-call, on AWS, Google Cloud or Cloudflare.
Yes. User research, flows, clickable prototypes, design systems and accessibility audits, designed with the engineers in the room so the designs ship as components rather than as a file nobody can build.
A fractional CTO is a senior technical leader who works part-time for a company, usually on a monthly retainer. Hidook offers it as part of its technical partnership: roadmaps, architecture reviews, technical due diligence, vendor management and hiring help.
Retrieval-augmented generation (RAG) over company documents and data, copilots inside existing products, agents that call tools and APIs, document and data extraction, and the evaluation, monitoring and cost control around them.
Every AI feature gets an evaluation set built from real questions and edge cases before it ships. Accuracy, groundedness, latency and cost per answer are measured from week one and re-run on every change, like tests.
Hidook works with Anthropic’s Claude models through the Claude API and other major model providers. The model is chosen per task by measuring candidates against the evaluation set for accuracy, latency and cost.
Usually RAG first: it keeps answers grounded in current company data and lets the system cite sources. Fine-tuning helps for a narrow, stable task with many examples, such as a fixed output format or tone, and is rarely the first step.
Hidook designs for the client’s data rules from the start: what may be sent to a model provider, where data is stored and processed (including EU data residency when required), who can see which documents, and full logging of what the model saw and answered.
Instafidget (https://instafidget.com) is a Hidook product that embeds a styled Instagram feed on any website. It offers 13 layouts, including grid, carousel, masonry, mosaic, slider, spotlight and timeline, works on WordPress, Webflow, Squarespace and plain HTML with one snippet, and connects to Instagram through the official API with read-only access.
Hidook Site Builder (https://lp.hidook.agency) is a Hidook product for local businesses. Owners pick a design made for their trade (restaurant, salon, trades, professional services or pastry shop), edit texts, photos, menu and colours in the browser with a live preview, and publish the website the same day.
Running products with real users keeps the team honest about the same things clients care about: uptime, performance, billing, support and the cost of every feature after launch.
Not on the list?
Ask us directly. A senior engineer replies within one working day.
Cookies, briefly.
With your OK we use Google Analytics to see which pages are useful. No ads and no selling data. Privacy