AI / From use case to operation
AI development company in Vietnam for practical business solutions
Friday Works is an AI development company registered in Ho Chi Minh City, Vietnam and operating fully remotely with local and international teams. We design AI solutions for business around a specific bottleneck, controllable data sources and measurable success criteria, beginning with a narrow pilot and expanding only when operational evidence is clear.
- Indicative budget
- 7–40 million VNDFor a pilot or first phase; AI model usage fees are separate
- Pilot timeline
- 1–6 weeksDepending on the problem, data readiness and integrations
- Control
- Human approvalAI suggests; an owner approves decisions with consequences
- Scale-up decision
- Based on measurementQuality, time and cost compared with the current way of working
Common problems
What businesses usually ask us to solve
Situation 01
Leadership wants to use AI but does not know which task to start with.
How Friday Works responds
We review repetitive, slow or error-prone work, score it by value and data readiness, and choose one problem worth piloting first.
Situation 02
You tried ChatGPT, but the answers ignore your company's documents.
How Friday Works responds
We use RAG, which lets the AI look up permissioned internal documents before answering, so every answer has a source users can check.
Situation 03
You worry that internal data will leak into AI tools.
How Friday Works responds
We define which data may be used, mask sensitive information, set access controls and choose an AI service plan that fits your data-retention requirements.
Situation 04
The AI demo looked impressive, but nobody can measure the real impact.
How Friday Works responds
Every pilot has a baseline, an evaluation set and clear metrics: quality, handling time, cost per task and the share of cases passed to a person.
Project types
Projects we typically take on
Internal knowledge assistants
Find and synthesise information from permissioned documents, with source references users can verify.
Sales and service assistance
Classify requests, prepare response drafts or suggest next actions within defined boundaries.
Content processing and review
Summarise, classify, extract or review large volumes of content with a managed exception workflow.
When should you start with AI?
A good fit when
- There is a repetitive, time-consuming task with a person accountable for the result.
- Input documents or data exist and you know who may access them.
- You accept starting with a narrow, measured pilot before expanding.
Not a fit when
- You want AI to make legally or financially consequential decisions without an approval step.
- You need a specific ROI commitment before a baseline and operating data exist.
Work in production
Real products serving real users
How we work
Step by step: who does what and what you receive
Example schedule for an AI pilot tied to one workflow, after the problem is chosen (3–6 weeks). Simple automation can be shorter; production rollout or expansion to other workflows is planned once pilot data exists.
- 01Before signing
Conversation & direction
Find the work that consumes the most time and check whether AI is genuinely the right answer.
Friday Works does
- Hear which tasks are repetitive, error-prone or labour-intensive.
- List candidate problems and make an initial assessment of the available data.
- Propose a narrow pilot with clear measures, or say plainly when AI is not a good fit.
Your part
Share the workflow, the workload and who owns the outcome.
You receive
- A pilot proposal: problem, data, success criteria and scope
- Schedule and quotation
CheckpointSign the proposal and pay the first instalment (40%).↗︎
- 02Week 1
Problem & data assessment
Measure today's process so the pilot can be compared fairly.
Friday Works does
- Interview the people who do the work and measure a baseline: time, errors and cost.
- Identify data sources, access rights and sensitive information.
- Create an evaluation set with correct answers confirmed by the business.
Your part
Provide documents and masked sample data, and confirm the reference answers.
You receive
- A problem and data-risk assessment
- Baseline and evaluation set
CheckpointAgree the pilot's success criteria.↗︎
- 03Weeks 1–3
Prototype
Test AI on real data before it enters the workflow.
Friday Works does
- Choose an AI model by quality, speed, cost and data requirements.
- Write model instructions, configure document retrieval and set guardrails.
- Run the evaluation set and compare against the baseline.
Your part
Review sample outputs and point out where the AI misreads the business.
You receive
- A prototype running on real data
- A first evaluation report
CheckpointResults demo and the second instalment (40%).↗︎
- 04Weeks 3–5
Pilot in the real workflow
Bring AI into the tools your team already uses, at small scale and with human approval.
Friday Works does
- Connect to the website, email, Google Sheets, CRM or existing software through APIs, webhooks or automation tools.
- Add human approval, activity logs and alerts.
- Hand cases to a person when the AI is not confident enough.
Your part
Assign pilot users and give weekly feedback.
You receive
- A pilot running in the real workflow
- A quality and cost tracker
- 05Weeks 5–6
Measure & decide
Use the numbers to decide whether to expand, adjust or stop.
Friday Works does
- Compare quality, time and cost with the baseline.
- Analyse errors, exceptions and cases passed to people.
- Estimate the monthly operating cost in production.
Your part
Review the data together and choose the next step.
You receive
- A pilot results report
- A recommendation to expand, adjust or stop
CheckpointPilot acceptance and the final instalment (20%).↗︎
- 06After the pilot
Handover & operation
You understand the configuration, the costs and how to run the solution.
Friday Works does
- Hand over source code, configuration, model instructions and accounts.
- Show how to update data and read logs and costs.
- Fix defects during the warranty period.
Your part
Name a person who monitors quality and cost each month.
You receive
- Operating documentation and accounts
- Three months of defect fixes for Friday Works' implementation
Deliverables
What you receive
Assessment & plan
- Prioritised AI opportunitiesEach with expected value, required data and risks.
- Baseline and evaluation setAn objective basis for comparing before and after AI.
- Data-risk assessmentWhich data is used, who may access it and what must be masked.
AI pilot
- Prototype and in-workflow pilotRuns on real data inside the tools the team already uses.
- Model and retrieval configurationModel instructions, knowledge sources and guardrails.
- Human approval stepsOwners approve important results; the AI hands over when unsure.
Measurement & decision
- Quality and cost trackerAccuracy, handling time, cost per task and hand-over rate.
- Pilot results reportCompared with the baseline, including error and exception analysis.
- Recommendation and roadmapExpand, adjust or stop, with an operating-cost estimate.
Handover
- Source code and configurationWith AI service accounts registered in the company's name.
- Operating documentationHow to update data, read logs and monitor costs.
- Three months of fixesDefects in Friday Works' implementation are fixed after acceptance.
Technology and tools we commonly use
The exact stack is chosen for each project and recorded in the proposal rather than forcing one platform onto every problem.
- AI modelsOpenAI, Anthropic Claude, Google Gemini or another suitable model, chosen for quality, speed, data and budget.
- RAG and vector databasesGround answers in your documents, products and policies, with references.
- Automation and integrationAPIs, webhooks, n8n, Make or custom flows connecting AI to existing tools.
- Control and monitoringPermissions, data boundaries, logs and usage-cost tracking.
Collaboration
What both sides prepare
What you prepare
- A process owner accountable for the pilot's outcome.
- Documents and sample data with sensitive details masked.
- Current figures for time, volume and errors, or permission to measure them in week one.
- AI service accounts in the company's name, or agreement on how usage is billed.
- Pilot users who give regular feedback.
What we check before handover
- The agreed quality threshold is met on the evaluation set.
- Results include a source or reason users can check.
- Uncertain cases go to a person instead of being guessed.
- Sensitive information does not appear in logs or outputs without authorisation.
- Cost per task is measured and stays within the expected budget.
- Every run is logged so it can be traced when needed.
Pricing
Indicative packages and how the budget is set
Related starting packages
Version 20 October 2026 · Unit: million VNDWorkflow automation
7–18million VND / workflow
Estimated: 1–3 weeks
Baseline scope. One workflow spanning up to two systems and ten steps; one approval where needed; notifications, logging and basic error handling.
Not included: Third-party platform or API usage fees.
Document-grounded AI chatbot
15–30million VND / first phase
Estimated: 3–5 weeks
Baseline scope. One website chatbot in one language; up to 100 machine-readable document pages; cited answers, missing-evidence notices and a document update screen.
Not included: Scanned-document OCR, Zalo OA, acting on behalf of users, model fine-tuning or AI usage fees.
Document extraction
15–35million VND / pilot
Estimated: 3–5 weeks
Baseline scope. Up to two document templates, fifteen fields per template and 100 test pages; PDF/image intake, extraction, review screen and Excel export.
Not included: Universal templates or handwriting by default; production-scale operation is quoted after the pilot.
Business research tool
20–40million VND / pilot
Estimated: 3–6 weeks
Baseline scope. One business cohort; pilot for up to 100 entities using public or authorised data; website verification, source capture, manual-review flags and Excel export.
Not included: Guaranteed personal emails/phone numbers, search/data fees or production operation.
These are indicative budgets for new engagements, not fixed quotations or prices from completed contracts. Scope, schedule and cost are confirmed for each project before work begins.
Typical timelines
- AI-assisted automation (A01)
- Often 1–3 weeks for one workflow of up to ten steps across two systems.
- Chatbot, document or research pilot (A02–A04)
- Often 3–6 weeks depending on data access and integration requirements.
- Production roadmap
- Planned after the pilot based on observed quality, adoption, cost and risk.
The budget depends on
- The number, format and readiness of data sources.
- Model choice, usage volume and data-retention requirements.
- The number of workflows, integrations and human approval points.
- Evaluation, security, observability and operational support requirements.
Proposed payment schedule
- 40%Project start
- 40%Demo & core features
- 20%Acceptance & handover
Before a quote is final
- Agree the page, feature, data and acceptance-criteria list before implementation.
- The schedule starts after scope approval, receipt of required inputs and the first payment.
- The proposal includes three months of defect correction for Friday Works' implementation after acceptance.
- Domain, hosting, email, AI/API, data, licences, third-party services and applicable tax are shown separately.
After handover
Warranty, maintenance and next steps
Three months of defect fixes
Defects in the work Friday Works delivered are fixed for three months after acceptance and are not charged as maintenance.
Monthly maintenance
2–5 million VND / month
Up to 6 hours/month · One application; log review, compatibility updates and technical support.
Support is provided during business hours, not as a 24/7 service. Infrastructure and work above the included allowance are quoted separately. Warranty defects are not charged as maintenance.Frequently asked questions
Before we begin
01What is an AI solution for business?
An AI solution for business introduces classification, extraction, knowledge retrieval, content generation or recommended actions into a business workflow with defined data, access controls and evaluation criteria. It may involve AI agents, RAG chatbots, Document AI/OCR, internal assistants or AI automation, but it should begin with a measurable business problem.
02Does Friday Works provide AI development from Vietnam?
Friday Works is registered in Ho Chi Minh City, Vietnam and operates fully remotely with local and international teams on focused AI pilots, internal knowledge assistants, workflow automation and production integrations. Scope begins with the business problem, available data, access controls and measurable acceptance criteria.
03What determines the cost of an enterprise AI implementation?
Enterprise AI implementation cost depends on the use case, data readiness, model and usage volume, system integrations, quality evaluation, guardrails, security and operational requirements. A focused pilot measures quality, handling time and cost per task before the business invests in production or broader rollout.
04Do we need perfect data before starting with AI?
A business does not need perfect data before starting with AI, but the team must know which data is trustworthy, who may access it and how output will be checked. A pilot can also reveal which data needs preparation before expansion.
05Do we need to train our own AI model?
Training your own AI model is not necessarily required. Many use cases work with existing models combined with instructions, retrieval and dedicated guardrails. The choice depends on quality, cost and control requirements.
06Does Friday Works guarantee ROI?
Friday Works does not promise an outcome without a baseline and operating evidence. The pilot measures quality, time, cost and risk so the business can make an informed decision.
07Will our data be used to train AI models?
Whether business data is used to train AI models depends on the provider's terms and the account type. Friday Works prefers API plans that do not use customer data for training by default, send only the data that is needed and mask sensitive information before processing.
08What will AI cost each month?
Monthly AI cost depends on the number of runs, document length and the chosen model. The pilot measures the real cost per task to estimate the monthly figure; model usage fees are separate from implementation and monitored monthly.
Talk to Friday Works
Find an AI use case worth testing
Start with the workflow consuming time, the data available and the result that needs to improve. We will help define a pilot with clear boundaries.
Describe your projectAfter you send a request
- 01Send a short brief
Your goal, current situation and preferred timing; attachments are welcome.
- 02A reply within one business day
Friday Works sends clarifying questions or proposes an online call.
- 03Discuss the scope
Review the goal, risks and an approach that fits the budget.
- 04Receive a written proposal
Scope, milestone schedule, cost and acceptance criteria.