Supply chain financing for informal businesses, powered by AI.

What makes your solution innovative?

Our main element of innovation has to do with our value proposition, which is made up of our own algorithm, based on the information we collect through our own platform. Likewise, our focus on productive loans for microenterprises, by serving a segment historically excluded from the financial system, preferential access for micro-entrepreneurs to supplies at discounted rates with affiliated businesses where they can use the "ladder credit" line; the use of artificial intelligence and automated learning to assess creditworthiness; and our offer of access to capital more quickly through our platform.

Our main asset is the solution developed from the diversity of information collected and the robustness of our credit scoring system. The information considered, among other parameters, the credit history of microenterprises, the reputational history, validation, and recommendation of peers, clients, and suppliers of the microenterprise. The analysis of this information, in conjunction with socioeconomic and regional information, allows us to create credit scores and thus open the doors to loans for microenterprises in the informal economy whose access to credit is very limited or non-existent. In our system, we have information about the person and their business, the products they sell, their reputation, and with this information the algorithm - already technically and functionally validated, currently in production - we can estimate the probability of repayment. The artificial intelligence system is based on algorithms known as random forest, regressions, and cross validations testing two types of algorithms. In addition, we have algorithms developed from deep neural networks, which, fed by photos and explanatory videos of the products marketed by microenterprises, allows us to analyze and interpret the truthfulness and quality of the products.

The data analyzed by the risk prediction algorithm are financial and non-financial. Among the non-financial data are identity validation, business videos of the applicant, product photos, a 48-question form (with validating crosses) about their business and income. We also request information from their trust network. In addition, we ask the applicant to refer clients and suppliers and we contact them through a chat with validating questions to verify the support of this network for the applicant. Among the financial data, we request access to transactional information in digital wallets through an open banking tool. This information allows us to understand the customer's debt capacity.

Our algorithm already has a training set of more than 9,000 credits and has reached an AUC level of 0.71, which exceeds market standards for credit risk assessment algorithms.

*The AUC is the area under the ROC curve (an acronym in English: Area Under the ROC Curve) This score gives us a good idea of how well the model is working. The ROC curve is a graphical representation of the sensitivity of a binary classifier system. A didactic example in https://aprendeia.com/curvas-roc-y-area-bajo-la-curva-auc-machine-learning/

Describe in simple terms how and why you expect your solution to have an impact on the problem.

Our solution allows the most vulnerable people to find a legal and fair financial means to maintain or increase their life chances. Our solution is not a social assistance system, but it does open opportunities for populations that are highly risky from the perspective of a traditional financial entity and even for government programs. They are excluded populations because there is no available information and the technological investment to obtain relevant data from this population and develop financial evaluation models is highly costly. Our solution is a path to connect knowledge, information and alternative risk assessment models in such a way that more and more people are served by the financial system. As we enable more capital with a vision of financial inclusion based on an assessment centered on the needs of micro-entrepreneurs, they can increase their livelihoods and depend less on informal lenders, obtaining higher profitability and improving their living conditions. Currently, 38% of our clients still access the informal lender. We hope that by 2030, less than 20% will access this financial option and manage to build a healthy and transparent credit history for other financial companies that provide other services (educational, housing, etc.). On the other hand, we have seen how our solution has managed to reach 32% of people who have never been served by the traditional financial system and 25% who have also not been served by the commercial system to pay for goods and services.

On the other hand, we know that entrepreneurs also need educational accompaniment and to open opportunities for their businesses. That is why we have specialized training for micro-entrepreneurs that we teach through WhatsApp and multiple benefits provided by our partners such as: brand creation, digital marketing, basic finance, CRM, advertising, presence in digital media, legal support, among many others.

Therefore, our greatest impact is in serving those who the traditional financial system does not know or is not interested in serving so that they can improve their goods and living conditions and, on the other hand, in helping them to build a credit history that serves to unlock other types of complementary financial services. In this way, our solution impacts relevant sustainability areas and cannot only be seen as a traditional credit service.

What are your impact goals for your solution and how are you measuring your progress towards them?

The indicators and impact goals that we continually monitor are both: 1) operational and management as well as 2) social impact.

The main management indicators are:

• Growth: 28% MoM (51,900 new clients, 16,907 renewals and $7,000,000 allocation)
• Default: 12% in the conventional channel, consistent by end of Dec (reducing the r&d portfolio)
• Engagement: 60% of the total potential recurring clients accept applying for a recurring credit.
• CAC: less than 2 dollars per credit disbursed.
• AUC algorithm: greater than 0.70. It is currently 0.71.
• NPS: greater than 90%. Today our NPS is 95%.

On the other hand, we follow important impact indicators to measure the achievement of our mission:

1. Gender inclusion: 70% of customers will be women in the active credit portfolio by 2025. Currently the indicator is 55%.
2. Reduction in the number of Quipu customers who continue to resort to informal lenders: reduction from 38% (2023) to 20% (2030).
3. Increase in the average monthly income of customers with credit in the active portfolio: that they go from having an income of US$250 to US$600 by 2025.
4. Financial inclusion: achieve 50% of Quipu customers have never had debts or history in the financial or commercial sector. Currently we have included 32% who had no history in the financial sector and 27% who did not have access to the commercial sector.

Additionally, we are preparing the following environmental impact indicators that are aligned with the Paris Agreement and which we hope to start implementing during 2024 and 2025:

1. We will systematically identify and manage climate-related risks in our portfolio.
2. We will ensure that loans are aligned with the Paris Agreement.
3. We will establish an Action Plan with key performance indicators and goals with specific dates related to the carbon footprint in our portfolio and with our entrepreneurs.
4. We will include in our educational modules one on environmental responsibility to train entrepreneurs in good environmental practices and green business creation.

Describe the core technology that powers your solution.

Our solution is built upon our AI-powered technology, a sophisticated system that incorporates our proprietary algorithm and utilizes the information collected on our platform. This technology is the heart of our operation, a system that guides us in our endeavour to provide productive loans for microenterprises, a sector traditionally overlooked by the conventional financial system. Our method of assessing creditworthiness is underpinned by artificial intelligence and machine learning. This technology allows us to provide rapid access to capital through our platform, a feature that sets us apart in the industry.

Our most valuable asset is our solution. This system, developed using diverse information and a robust credit scoring system, considers an array of parameters. These parameters include the credit history of microenterprises, reputational history, and validations and recommendations from peers, customers, and suppliers. We don't just stop at these traditional metrics. By analyzing this information in conjunction with socioeconomic and regional data, we are able to generate comprehensive credit scores.

Our algorithm utilizes this information to estimate the probability of repayment, a prediction that has proven to be remarkably accurate. In our quest to perfect our credit scoring system, we employ known algorithms such as random forest and regressions. We also conduct cross-validation tests and use deep neural network algorithms. These neural network algorithms are fed with photos and explanatory videos of the products sold by microenterprises. This allows us to analyze the authenticity and quality of the products, a level of scrutiny that ensures we only support genuine businesses.

Our risk prediction algorithm doesn't just focus on financial data. It also analyzes non-financial data that includes identity validation, business videos, product photos, and a comprehensive 48-question form about the applicant's business and income.

Our algorithm, which has been trained on a dataset of over 9,000 credits, has achieved an AUC level of 0.71. This performance not only validates our system but also surpasses market standards for credit risk assessment algorithms.

In conclusion, our Quipu Score allows informal businesses to build credit history and access productive loans even if they have a bad report in the credit bureau. We have built diverse AI models based on proprietary data:

• Video and image score: we use deep learning to analyze videos and images of the businesses.
• Validation questions about their business: type of business, location, sales etc.
• Referrals and network score: we validate their network of clients and suppliers
• Digital footprint score: we use +80,000 metadata points on how clients use their phone, comparing to +6M productive loans of the global south
• KYC score: we use their identity validation data to build a score
• Banks and wallets accounts transactions score

Which of the following categories best describes your solution?

A new application of an existing technology

Please select the technologies currently used in your solution:

• Artificial Intelligence / Machine Learning
• Software and Mobile Applications

If your solution has a website or an app, provide the links here:

quipu.com.co

In which countries do you currently operate?

• Colombia

Which, if any, additional countries will you be operating in within the next year?

• Mexico

How many people work on your solution team?

We are 45 people working at Quipu and we are located in more than 5 countries: Colombia, Argentina, Peru, Ecuador, and the United States.

Full-time staff: 43

Contractors: 2

How long have you been working on your solution?

We have been working on our solution for over 3 years. The first year, we iterated and developed the solution. The second year, we stabilized our proprietary algorithm based on AI. In the last year, we evolved our solution and grew our user and customer base.

Tell us about how you ensure that your team is diverse, minimizes barriers to opportunity for staff, and provides a welcoming and inclusive environment for all team members.

At Quipu, we believe in the diversity and equity of our team. Our two co-founders are women and have always fought for gender equality at Quipu. Currently, 55% of employees are women and 45% are men. But they are not only interested in equality in terms of gender balance but also in inclusive practices. This is why at Quipu we have a policy of flexibilities where explicit support is given to pregnant women, in their medical check-ups and in everything related to their pregnancy process. Likewise, there is flexibility with those who are already parents so that they have spaces with their children when they need them.

And although at Quipu we already have this policy, since last year we have been working with Promujer and Lefil to make Quipu a more equitable and diverse work environment, especially in terms of gender. For more than 10 months we have received support from the "Gender Perspective Integration" program and we have set ourselves the task of eliminating unconscious biases in our work team in relation to diversity. This is why we have had mentoring and practical sessions subsidized by Promujer with an external consultant called Pink.

Today at Quipu, 10% of employees are LGBTQ+ and this percentage is expected to increase as we continue to grow, including transgender people.

What is your business model?

Our business model consists of offering a revolving line of credit that increases the amount as users pay their loan well and use the platform. We provide the possibility of obtaining the loan in any digital wallet that is in the name of the entrepreneur.

Regarding the credit we approve, we make several charges:

• Interest: 33% currently. We always abide by the usury rate of consumer credits in Colombia.
• FGA: 25%. We charge this percentage to guarantee a part of the debt with the National Guarantee Fund (a Colombian public entity) and to not require collateral or a co-debtor from the entrepreneur.
• Digital instrumentation: 10 USD on all credit.

We divide all these costs into the installments of each loan depending on the approved value. Our first loan is usually 125 USD.

Currently, we have three ways of reaching users:

(i) Advertisements and publicity on social networks, the cost per lead on social networks is less than US$0.3 per applicant and Quipu can receive up to 500 applications a day, this first channel is effective for digital and mass user arrival.

(ii) Alliances: with active alliances such as Claro, Bancolombia, Bancolombia a la mano and Nequi we estimate that we can reach more than 30M users in the country. In these alliances, we offer our productive credit to all their customers in Colombia (ads, push notifications, SMS, emails). Those who want to apply for the loan must enter the channel provided by each ally. With some, we have a revenue share model where we share success and a part of the profits, and in others, they are referral models to mobilize the digital channel more.

(iii) MGM (member get member): Management and promotion of networks and connections among users, and attraction of referrals. We aim for this channel to represent 20% of our acquisition channels by the end of 2024.

Our model is directly aimed at individuals with some type of entrepreneurship. Our B2C model is: 

Approved customers pay a 25% guarantee of their credit to open their credit line (this payment goes to the National Guarantee Fund), the guarantee in Colombia works as a 'co-debtor' and allows customers who do not have other support, to access credit. The customer also pays an administrative fee called "digital instrumentation" for the value of 10 USD and, in addition to the credit, can access the business showcase of the platform and exclusive benefits for Quipu customers, as well as to carry out the credit process digitally. Subscription is optional, the customer can choose not to pay the subscription and carry out the credit application process physically. The first loan is US$125 and increases up to US$1000, the term varies between 5 and 24 months depending on the cash flow of the business and the customer's payment capacity. The annual effective rate of Quipu loans is 33%, while the official maximum rate is 34% per year, and the rate of informal lenders is between 400% and 600% per year.

Do you primarily provide products or services directly to individuals, to other organizations, or to the government?

What is your plan for becoming financially sustainable, and what evidence can you provide that this plan has been successful so far?

Our experience in credit placement began in early 2022. Throughout the process, we have experimented with the information collected and the type of approval, which has made us decrease the default (-13% anual) while controlling the customer base.

To be sustainable, we have opened several investment rounds, where the last one was 2.5M USD led by Grupo Bancolombia (1.5 USD) and BID Lab (750 USD). This seed round has given us the possibility of having a runway of 24 months before opening the next round (Series A).

Currently, we have institutional debt funds (SIVO + Addem Capital) with whom we have approved lines of up to 5M USD at a cost of less than 20% per year.

To achieve our sustainability we know that by 2024 we must reach a total of 80k customers with disbursed credit and with a recurrence of 80%. Likewise, we must continue diversifying our product to make loans that start at 500 USD and rise to 5k USD (micro-credits).

On the other hand, we are frequently looking for grants that boost our mission and help us evolve our products. We were recently chosen by Open AI and have been awarded by Fundación Cartier, Mastercard, Visa, and others who have believed in our mission.