What governments can improve
Clarify and simplify procedures, digitise services and improve regulations.
Draft product definition · UNCTAD
Needs
Assessment
AI changes what institutions need and how UNCTAD can assess and support them. From questionnaires to tailored country and regional assistance proposals for donors.
First application: free and special economic zones in Central Asia. Designed to be configured for other regions, fields, institutions and procedure families.
Three areas governments can improveThe needs assessment records what each government has and what it wants to improve in these three areas.
Clarify and simplify proceduresMake procedures clear and transparent, and remove unnecessary steps.
Digitise servicesProvide public services online through procedures that have first been clarified and simplified.
Improve regulationsWrite clear regulations that reflect how the improved procedures should work.
How governments use the needs assessment
Government institutions and UNCTAD assess the needs together. The findings define possible solutions and produce documents for internal reform decisions and donor requests.
How governments assess their needs
A regional or sectoral partner can organise the assessment for its members. At country level, the presidency or a lead ministry can organise it for participating institutions. The organiser, participants and UNCTAD use the tool together. A scoping questionnaire is followed, where needed, by a detailed assessment that the government validates.
Solutions and UNCTAD assistance
The findings define the problem and guide the choice of solution. This part contains the UNCTAD assistance menu, including tailored support, configurable templates and the OpenGov approach.
Documents for reform and funding
The process produces the documents governments need to build internal support for reform and request country or regional funding from donors.
How AI changes needs assessment and assistance
AI is changing what institutions can do, what users expect, how work is organised and how assistance can be designed and developed. A needs assessment that records only today’s position may prepare an institution for a world that has already changed.
How AI changes the assessment
Reusable documents and AI-assisted processing can accelerate questionnaire design, evidence organisation, comparison and drafting. This makes broader and more regular assessment possible at substantially lower cost, with expert validation retained.
How AI changes institutional needs
Institutions may need different services, procedures, staff roles, data, systems, rules and safeguards. The assessment must examine both current gaps and the changes likely to reshape the institution’s work.
How AI changes UNCTAD assistance
AI techniques can help clarify and simplify procedures, specify and develop digital solutions, and expose ambiguities in regulations. The tools and methods must keep evolving as AI capabilities improve.
Each assessment covers present needs and expected changes.
Each assessment examines how AI affects the three assistance areas. UNCTAD uses AI to support the work. Participating institutions decide policy and approve the results.OpenGov and the agentic direction
Governments publish their rules in forms that people can understand and machines can use. Agentic services read those rules, reuse authorised data and ask for what is missing to help people understand requirements and apply. Agents may also support officials, but the competent authority validates and decides. Open standards allow governments, communities and providers to keep improving the tools.
OpenGov
Mission and expectations
Services and procedures
Data and systems
People and organisation
Governance and trust
Appropriate response
How AI may change the institution’s mandate, priorities, users’ expectations and the public value it is expected to deliver.
Where AI may support clearer information, faster service delivery, better decisions or new forms of assistance, and where ordinary simplification or digitisation remains the better response.
Whether reliable data, interoperable systems, infrastructure and information-management practices can support the possible solutions.
Which skills, roles, work practices, leadership and change-management capacity will be needed as tasks and services evolve.
What legal authority, accountability, human oversight, privacy, security, fairness and transparency safeguards are required.
Which needs call for AI-enabled assistance, which call for another digital or regulatory response and which require further investigation before a solution is chosen.
What AI can accelerate
Configure questionnaires and documents, organise and compare responses, extract cited evidence, flag gaps and inconsistencies, and draft profiles, solution specifications, prototypes, reports and proposals from approved source material.
What UNCTAD and participating institutions must decide
Define the method and assistance products, validate scope and evidence, exercise technical and legal judgement, approve findings, specifications and recommendations, and lead country and donor engagement.
How AI lowers the cost of needs assessment
Needs assessment has often required dedicated donor financing before implementation can begin. Reusable questionnaires, document templates and AI-assisted processing can make that preliminary work much faster and less costly.
The proposed model lowers the cost of the first assessment and reserves more project funding for assistance.
The model is designed to give UNCTAD a lower-cost assessment front end and use external financing where it adds the most value: validating complex needs, designing solutions and delivering assistance. Assessment remains proportionate to the decision it must support.
Two levels of needs assessment
The work proceeds in stages. A broad questionnaire identifies possible needs across many institutions. A detailed assessment then validates the needs of a selected country or institution. UNCTAD specifies the solution and prepares a project only after the detailed assessment.
Scoping needs assessment
A broad, rapid questionnaire across many respondents establishes a self-reported picture of where institutions stand, what they need, how AI may change their operating context and which assistance areas interest them.
Output: respondent profiles, a consolidated evidence map, initial priorities and candidates for follow-up.
Proposed resource route: a lower-cost front end, potentially supported through the regular programme of work.
Detailed individual assessment
Focused follow-up with a country or institution adds documents, interviews and evidence on procedures, systems, institutions, legal constraints and readiness for an AI-transformed environment.
Output: validated needs, recommendations and evidence detailed enough to specify an appropriate solution.
Proposed resource route: regular programme of work where approved and feasible, or a smaller preparation budget.
Solution specification and project preparation
Once a need is validated, go deeper: specify the rules, data, functions, roles, interfaces, safeguards and interoperability the solution requires, then define ownership, activities, results, sustainability, timetable and budget.
Output: an open, modular solution specification that can later support conventional or AI-assisted development, plus a draft country, regional or combined proposal and implementation plan for donor consideration.
Proposed resource route: project preparation or donor financing linked to delivery.
Delivery and cost guardrail: this is a proposed model, not a funding commitment. Its regular-programme and donor-funded boundaries require institutional approval. Benchmark staff time, processing cost and validation effort before making a public claim such as “almost free.” Lower cost must not mean weaker validation, data protection or institutional ownership.
The needs assessment
A short, broad assessment identifies the respondent’s current position, expressed needs and interest in the three assistance areas. It also captures what is already underway, how AI is changing the operating context, who is involved and what kind of follow-up may be useful.
A regional or sectoral partner can convene its members. At country level, the presidency or a lead ministry can coordinate the participating institutions. The assessment also identifies who validates the findings.
Who is responding, which institution or territory the answers cover, and whether the response represents one entity, several entities or a national system.
What frameworks, procedures and systems exist; whether the priority procedure already exists or must be created; how transparent, simplified and digital current procedures are; and what reforms are underway.
Ask first about problems and desired outcomes in neutral terms. Then ask which of the three assistance areas, products or delivery tracks interest the respondent.
What should be clarified first, what should be simplified or digitised first and how urgent the selected priority is.
Which competent authorities must participate, what capacity and systems already exist, and what further validation is required.
How AI may change the institution’s mission, services, procedures, skills, data, systems and safeguards, without assuming that AI is necessarily the appropriate solution.
Whether regional, country-specific, institution-specific or combined follow-up would be useful, and which partners or donors are already involved.
Institution-specific procedures
Procedures governing a user’s legal or operational relationship with the institution being assessed.
- Admission, approval or registration
- Institution-specific licences and certificates
- Renewals, reporting and status changes
Wider ecosystem procedures
Procedures that users may need from other authorities in order to establish, operate, trade, work or comply.
- Company, tax and employment registrations
- Import, export and customs procedures
- Sectoral, construction or environmental approvals
Solutions and UNCTAD assistance
The detailed assessment defines the problem. The government and UNCTAD compare assistance options in the three areas and agree on a preferred solution. Each option can be tailored or based on a configurable template.
Improve or create procedures and digital services
Use the evolving eRegulations and eRegistrations concepts to improve existing procedures or design new ones, digitise services and improve the regulations that govern them.
OpenGov
Publish rules in forms that people can understand and machines can use. Agentic services use the rules and authorised data to help people apply, while the competent authority validates and decides.
When a procedure already exists
Document how it works, simplify it, digitise the improved procedure and use the result to revise the regulations. This remains a sound method.
When a procedure must be created
A new law or institution may require a new procedure. Define the result, users, responsibilities, data, proofs, decisions and steps before implementation.
AI-assisted eRegulations
UNCTAD is developing an AI-assisted version of eRegulations to speed up procedure documentation, identify gaps and duplication, and suggest simplification for review by the responsible institutions.
SmartRules and eRnext
SmartRules structures the requirements and decision rules. eRnext uses them to guide applicants and generate digital interactions. Together they can shorten the time needed to simplify a service and implement the improved process.
01
Clarify and simplify procedures
Existing procedure. Document it step by step from the user’s point of view, including requirements, forms, cost, time, legal basis and responsible authorities.
New procedure. If a new law or institution requires a procedure, design the result, users, responsibilities, data, proofs, decisions and steps before implementation.
Simplify. Remove unnecessary steps, requirements and interactions, and clarify responsibilities before automation begins.
AI-assisted eRegulations. Extract procedure information from laws, forms, documents and interviews, prepare a first structured draft, identify gaps and duplication, and suggest simplification for institutional review.
Tailored assistance
Document and simplify the procedures selected by the government, using its institutions, rules and current practice.
Configurable template
Configure a standard procedure-publication structure, required data fields and simplification checklist for the selected procedures.
02
Digitise services
Digitisation follows clarification and practical simplification. For a new procedure, the approved design provides the starting point for the digital service.
Tailored assistance
Apply the evolving eRegistrations concept to automate selected procedures in their national and institutional context, with pragmatic simplification, agentic assistance and integration where required.
Configurable template
Configure a standard digital system with automated procedures, modules, branding, roles and data for the country.
SmartRules and eRnext. Structure requirements and decision rules, use them to generate forms and guided interactions, and help applicants complete the procedure. This can shorten the time between simplification and implementation.
03
Improve regulations
For an existing procedure, use the clarified, simplified and digitised process as evidence for better legal rules. When a new law or institution requires a new procedure, procedure design and legal drafting progress together, and the final rules reflect the approved process.
AI contribution. Extract draft machine-readable rules from existing texts, compare them with the optimised procedure and surface ambiguity, gaps and conflicting provisions for legal and policy review.
The reusable product is built around regulatory guidelines and good drafting principles. Model clauses or templates may support the work, but they do not replace adaptation to the country’s legal system.
The optimised procedure is evidence, not the sole source of law. Legal work must also address policy, powers, rights, remedies, sanctions, institutional mandates, constitutional and treaty constraints, consultation and the country’s adoption process.
Tailored assistance
Review existing rules or draft new ones with the government and national legal experts.
Configurable template
Adapt standard regulatory guidelines, structures and model clauses to the country’s legal system and policy choices.
Existing procedures are documented and simplified before digitisation. New procedures are designed before implementation.
Documents produced by the assessment
The process produces documents for respondents, government decision-makers, reform teams and donors. Each document uses the same validated evidence.
A readable record of each respondent’s scope, answers, priorities, current initiatives and open questions.
A brief or presentation that explains the problem, evidence, preferred option and decisions required inside government.
Validated findings, common needs, selected priorities, coverage and limitations for one country, sector or regional programme.
A concept note, presentation and project proposal that use the same evidence to request support for the selected reform.
The questionnaire identifies possible needs. A detailed diagnostic is required before UNCTAD recommends assistance.
Automated processing may organise evidence and identify follow-up questions. It must not create a fixed roadmap, legal recommendation or cost estimate from broad survey answers alone.Project and donor routes
The assessment supports several funding formats. Donors can choose a shared regional programme, selected country projects or a portfolio combining both.
| Route | Evidence base | What it can fund | Required before commitment |
|---|---|---|---|
| Regional or multi-country project | Consolidated patterns across participating institutions and countries, with coverage disclosed. | Shared methods, platforms, guidelines, capacity building, regional coordination and a country-support window. | Donor interest, programme governance, selection method, delivery capacity and validation of regional components. |
| Country-specific project | Validated national evidence and a deeper diagnostic with competent authorities. | A tailored combination of clarification, simplification, digitisation and regulatory work. | Country ownership, agreed scope, implementation partners, results, timetable and budget. |
| Combined portfolio | A regional core with validated country annexes or project concepts. | The entire programme, common components, selected countries or specific assistance tracks. | Clear separation of common and country costs, beneficiary-selection rules and coordination arrangements. |
Expression of needAssessment response and stated interest.
EligibilityScope, mandate, country ownership and fit with funding.
Deeper diagnosticEvidence, authorities, procedure ownership and readiness.
SelectionTransparent criteria applied to available resources.
Project agreementActivities, results, roles, timetable and budget confirmed.
Reusable product architecture
The marketplace product contains a stable core and configurable regional or sectoral instances. New instances adapt facts and context without rebuilding the method.
| Component | Reusable core | Configured for each instance |
|---|---|---|
| Product definition | Method, three assistance areas, AI-context lens, evidence rules, project routes and quality gates. | Field, institutions, geography, partners, programme name and product emphasis. |
| Public website | Information architecture, assessment explanation, action flow, accessibility and speaker integration. | Written copy, branding, languages, examples, partners, questionnaire link and speaker identity. |
| Questionnaire | Question families, conditional logic, scope and mandate, procedure categories, changing AI context, readiness and follow-up design. | Specific institutions, procedures, legal framework, AI-related changes, assistance products, terminology and countries. |
| Evidence processing | Participant profile, validation states, aggregation rules, coverage disclosures and follow-up questions. | Indicators, comparisons, report language, respondent types and approval route. |
| Donor package | Regional, country and combined project structures; beneficiary-selection method; results framework prompts. | Evidence, beneficiaries, activities, delivery arrangements, donors, budget and timetable. |
| Assistance products | Tailored methods and configurable templates for all three assistance areas, including the evolving eRegulations and eRegistrations concepts, OpenGov principles and regulatory guidelines. | Priority procedures, rules, data, integrations, system modules, AI opportunities, legal work and implementation sequence. |
| AI and skills | Product skill, content workflow, source-of-truth rules, review protocol and speaker derivation method. | Instance sources, approved terminology, written content, speaker text and implementation instructions. |
How to build an instance
Work backwards from the decisions and documents the assessment must support. Do not begin by writing the website or copying the questionnaire from another region.
Define the field, actors and changing context
Specify the institutions, users, geography, procedure families, competent authorities, population, invitation or sampling plan and who may speak for each level. Identify how AI may be changing the field and the institution’s operating environment.
Define the assistance products
State exactly what UNCTAD can offer in the three ordered areas, including AI-enabled techniques, delivery tracks, limits and required capabilities. Treat tools as evolving implementations, with open specifications as the durable output.
Design the evidence chain
Identify what participant, country, regional and donor documents must say, then list the evidence each statement requires.
Build the questionnaire and processor together
Every question must serve a defined output. Define consent, access, retention, correction, attachments and publication rules. Processing organises evidence and uncertainty; it does not invent recommendations.
Prototype the outputs
Render a respondent profile, internal reform brief, country or regional report, solution-specification outline and donor submission with synthetic test data before launch.
Write the public source of truth
Explain the assessment, assistance areas, outputs, project routes and participation conditions in the instance website.
Derive the speaker
Transform approved written content into concise spoken language. The speaker does not introduce additional facts, promises or qualifications.
Review adversarially and pilot
Run cognitive and translation pilots. Test representativeness, non-response, data quality, evidence claims, question interpretation, output logic, legal meaning, privacy and mobile usability.
Launch, learn and update the product
Run the instance, analyse real responses, improve the reusable core and capture validated lessons in the product skill.
Product quality gates
An instance advances only when the evidence and authority required by the next stage exist. These gates prevent a short questionnaire from becoming an unsupported diagnosis or donor promise.
Approve the purpose, three ordered assistance areas, product boundaries, stages, roles and permitted claims.
Define the unit and population, invitation method, respondents, legal and AI context, terminology, assistance variants, outputs and validation authorities.
Map every intended output claim to evidence. Ask neutral needs before product interest; capture scope, mandate, changing operating context and consent; separate procedure families; pilot language and interpretation; define access, retention, correction and secure attachments.
Deduplicate units, check completeness and contradictions, triangulate documents where required, disclose coverage and non-response, and let respondents correct their profiles. Country findings require competent national validation.
No solution specification, roadmap, activity package, cost, timetable or legal recommendation until a deeper diagnostic confirms priority, ownership, authority, feasibility, dependencies, safeguards and complementarity with existing partners.
Approve the written source first. Publish identifiable or country-level material only under agreed consent and validation rules. Spoken text renders the written source without adding policy, promises or qualifications.
Protocol for AI collaborators
- Read this product definition before reviewing or building an instance.
- Start from the approved assistance products and evidence requirements, not from existing page copy.
- Distinguish a broad assessment from a deeper diagnostic, recommendation or roadmap.
- Assess both present needs and the changes AI may bring. Treat AI as context and a possible capability, not as the default solution.
- Apply AI across the three assistance methods, but keep evidence, specifications, legal judgement and institutional decisions distinguishable.
- Never infer a country position from one institution or zone response without validated authority and coverage.
- Keep the ordered method visible: clarify and simplify, digitise, then improve regulations.
- Separate institution-specific procedures from wider ecosystem procedures and identify competent authorities.
- Write the public source of truth first; derive speaker text only after the written content is approved.
- Separate generic product decisions from instance facts. Do not carry Central Asia assumptions into another region.
- Surface open decisions and evidence gaps. Do not fill them with plausible invention.
Central Asia, the first instance
The World Free Zones Organization Regional Office for Central Asia organises the first application for zones in the region. UNCTAD provides the needs-assessment method and technical support.
- Reorder the proposal around the three assistance areas and place regulations last.
- Align the broad assessment with the three products and two digitisation tracks.
- Distinguish zone-specific procedures from wider business procedures.
- Add scope, mandate, readiness, desired outcomes and preferred project route.
- Ask how AI may change zone services, procedures, roles, data and safeguards without presenting AI as the default response.
- Replace the automatic fixed roadmap with evidence profiles and follow-up questions.
- Define regional, country and combined donor-project documents and beneficiary selection.
- Approve the written website before deriving Gulnura’s spoken text and captions.
Open product decisions
These decisions belong to the reusable product. They should be resolved before it is converted into a skill or offered through the marketplace.
Define their enduring concepts and methods separately from any current software implementation. Complete the AI-assisted eRegulations method for documentation and simplification, and define how SmartRules, eRnext, machine-readable rules, agentic assistance and open standards shorten reform design and implementation.
Define the standard procedure templates, modules, roles, configuration boundaries, hosting model, integrations and what remains genuinely ready-made.
Define UNCTAD’s method, legal expertise, guidelines, quality assurance, country adaptation process and the boundary between advisory support and national lawmaking.
Define who validates needs, selection criteria, decision authority, conflict management, transparency, country ownership and how donors participate.
Define the product owner, maintained assets, versioning, regional configuration process, support model and how lessons from instances update the reusable core.
Define the minimum AI-context questions for every instance, the deeper readiness evidence required before recommending an AI-enabled response, how AI supports each of the three methods and which evolving AI-related assistance products UNCTAD is prepared to offer.
Benchmark the staff time, AI-processing cost and expert-validation effort at each assessment level. Define which activities UNCTAD can provide through regular-budget work and which require project-preparation or donor financing.