A warning can be accurate and still arrive too late.
Someone receives a job offer through a message. The salary sounds unusually generous. The employer wants a quick answer. The person is hopeful, under pressure, or both.
In that moment, good advice may exist somewhere else: on a government website, inside a long guide, or behind the requirement to download another application. Reaching it takes time, attention and confidence—the very things urgency can reduce.
A Mexican project called Ramona AI offers a useful example of a different design choice. In September 2026, the International Telecommunication Union reported that Ramona AI had won the Digital Pioneers category at the UN-led Digital@UNGA Awards. According to ITU, users can forward a suspicious job offer through WhatsApp or Telegram and receive a risk assessment that explains warning signs. The service does not require an application download, registration or payment.
ITU says the system was informed by 200 interviews with survivors of labour trafficking and sexual exploitation. It had analysed more than 127,000 job offers across five Latin American countries since 2022, and more than 800 people had reported that it stopped them from responding to offers linked to trafficking.
These are figures reported by ITU from the initiative; the award announcement does not provide an independent evaluation of every assessment or outcome. The larger design lesson is an Evolv interpretation.
Put help inside the moment of need
Leaders often measure whether support exists. People experience whether support is reachable.
Imagine an organisation that creates an excellent guide for identifying fraudulent requests. The guide is technically available, but it sits six clicks from the homepage. It uses unfamiliar language and assumes that a worried person will calmly compare ten indicators.
The organisation has produced information. It has not necessarily produced usable protection.
Ramona AI’s reported approach begins in channels people already use. That choice reduces several small obstacles between concern and action.
The principle can travel beyond digital safety: when someone needs help, examine every step required to reach it.
Run a five-minute friction test
Choose one service your team says is available: reporting harassment, requesting a refund, asking for technical support, finding emergency information, or raising a safeguarding concern.
Start at the exact point where a person would first feel uncertain. Then count what they must do before receiving useful guidance:
- How many screens, transfers or searches are required?
- Must they create an account or provide information before understanding the options?
- Does the language make sense to someone unfamiliar with your organisation?
- Can they use the service privately on a basic device or a slow connection?
- What happens if their first description is incomplete?
This is an editorial exercise, not a validated research method. Its purpose is to expose avoidable effort.
Do not celebrate a lower click count by itself. A shorter route must still protect privacy, provide accurate guidance and connect people to qualified human help when a situation exceeds what an automated tool can safely handle.
Design for the pressured version of a person
Many processes are tested by people who already know how the system works. Real users may arrive tired, afraid, embarrassed, rushed or unsure whether their concern is serious enough.
That difference matters.
Consider a hypothetical employee who notices a suspicious request involving company funds. The reporting page asks them to classify the incident using internal terminology. They leave because they do not know which category applies.
A better opening might ask what happened in ordinary language, then guide the person to the appropriate route. The system can gather detail progressively instead of demanding perfect understanding at the beginning.
The goal is not to remove every question. It is to avoid making expertise the price of entry.
Keep explanation beside the signal
A red warning without an explanation can create dependence. A clear explanation can help someone recognise a similar risk next time.
ITU describes Ramona AI as providing a risk assessment that explains warning signs. For leaders, that suggests a useful standard: when a system recommends caution, show the user what triggered it in language they can understand.
Automated assessments can be wrong. An explanation gives the person something to examine, question and discuss. It also makes it easier to identify a poor rule or an unfair pattern.
For high-stakes situations, the next step should remain visible: a trusted authority, trained specialist or verified reporting channel. A digital tool can help someone pause; it should not imply that software alone has resolved the danger.
Ask where help currently waits
At your next service review, avoid beginning with “What new feature should we build?”
Begin with: “Where does our help currently wait, and what must a person overcome to reach it?”
The answer may reveal that the most valuable improvement is not a larger platform. It may be a clearer message, fewer required fields, an accessible channel, or a better handoff to a person.
The positive part of Ramona AI’s recognition is larger than the use of artificial intelligence. It is the decision to meet a risky message with an accessible way to pause and examine it.
Good support does more than exist.
It arrives close enough to be used.
Which service in your organisation would become more valuable if it moved one step closer to the moment people need it?
Editorial note: Prepared with AI assistance from the primary sources below. Evolv did not interview Ramona AI, its users or survivors. The organisational examples and friction test are original editorial proposals. Award recognition and self-reported outcomes should not be read as an independent validation of diagnostic accuracy or guaranteed protection.
Sources: ITU: Digital@UNGA Awards—Projects reaching people, 22 September 2026. Mexico City Secretariat of Citizen Security: warning about false job offers, 30 September 2026.
A warning can be accurate and still arrive too late.
Someone receives a job offer through a message. The salary sounds unusually generous. The employer wants a quick answer. The person is hopeful, under pressure, or both.
In that moment, good advice may exist somewhere else: on a government website, inside a long guide, or behind the requirement to download another application. Reaching it takes time, attention and confidence—the very things urgency can reduce.
A Mexican project called Ramona AI offers a useful example of a different design choice. In September 2026, the International Telecommunication Union reported that Ramona AI had won the Digital Pioneers category at the UN-led Digital@UNGA Awards. According to ITU, users can forward a suspicious job offer through WhatsApp or Telegram and receive a risk assessment that explains warning signs. The service does not require an application download, registration or payment.
ITU says the system was informed by 200 interviews with survivors of labour trafficking and sexual exploitation. It had analysed more than 127,000 job offers across five Latin American countries since 2022, and more than 800 people had reported that it stopped them from responding to offers linked to trafficking.
These are figures reported by ITU from the initiative; the award announcement does not provide an independent evaluation of every assessment or outcome. The larger design lesson is an Evolv interpretation.
Put help inside the moment of need
Leaders often measure whether support exists. People experience whether support is reachable.
Imagine an organisation that creates an excellent guide for identifying fraudulent requests. The guide is technically available, but it sits six clicks from the homepage. It uses unfamiliar language and assumes that a worried person will calmly compare ten indicators.
The organisation has produced information. It has not necessarily produced usable protection.
Ramona AI’s reported approach begins in channels people already use. That choice reduces several small obstacles between concern and action.
The principle can travel beyond digital safety: when someone needs help, examine every step required to reach it.
Run a five-minute friction test
Choose one service your team says is available: reporting harassment, requesting a refund, asking for technical support, finding emergency information, or raising a safeguarding concern.
Start at the exact point where a person would first feel uncertain. Then count what they must do before receiving useful guidance:
- How many screens, transfers or searches are required?
- Must they create an account or provide information before understanding the options?
- Does the language make sense to someone unfamiliar with your organisation?
- Can they use the service privately on a basic device or a slow connection?
- What happens if their first description is incomplete?
This is an editorial exercise, not a validated research method. Its purpose is to expose avoidable effort.
Do not celebrate a lower click count by itself. A shorter route must still protect privacy, provide accurate guidance and connect people to qualified human help when a situation exceeds what an automated tool can safely handle.
Design for the pressured version of a person
Many processes are tested by people who already know how the system works. Real users may arrive tired, afraid, embarrassed, rushed or unsure whether their concern is serious enough.
That difference matters.
Consider a hypothetical employee who notices a suspicious request involving company funds. The reporting page asks them to classify the incident using internal terminology. They leave because they do not know which category applies.
A better opening might ask what happened in ordinary language, then guide the person to the appropriate route. The system can gather detail progressively instead of demanding perfect understanding at the beginning.
The goal is not to remove every question. It is to avoid making expertise the price of entry.
Keep explanation beside the signal
A red warning without an explanation can create dependence. A clear explanation can help someone recognise a similar risk next time.
ITU describes Ramona AI as providing a risk assessment that explains warning signs. For leaders, that suggests a useful standard: when a system recommends caution, show the user what triggered it in language they can understand.
Automated assessments can be wrong. An explanation gives the person something to examine, question and discuss. It also makes it easier to identify a poor rule or an unfair pattern.
For high-stakes situations, the next step should remain visible: a trusted authority, trained specialist or verified reporting channel. A digital tool can help someone pause; it should not imply that software alone has resolved the danger.
Ask where help currently waits
At your next service review, avoid beginning with “What new feature should we build?”
Begin with: “Where does our help currently wait, and what must a person overcome to reach it?”
The answer may reveal that the most valuable improvement is not a larger platform. It may be a clearer message, fewer required fields, an accessible channel, or a better handoff to a person.
The positive part of Ramona AI’s recognition is larger than the use of artificial intelligence. It is the decision to meet a risky message with an accessible way to pause and examine it.
Good support does more than exist.
It arrives close enough to be used.
Which service in your organisation would become more valuable if it moved one step closer to the moment people need it?
Editorial note: Prepared with AI assistance from the primary sources below. Evolv did not interview Ramona AI, its users or survivors. The organisational examples and friction test are original editorial proposals. Award recognition and self-reported outcomes should not be read as an independent validation of diagnostic accuracy or guaranteed protection.
Sources: ITU: Digital@UNGA Awards—Projects reaching people, 22 September 2026. Mexico City Secretariat of Citizen Security: warning about false job offers, 30 September 2026.
No comments:
Post a Comment