Fintelite adalah perusahaan rintisan yang bertekad meningkatkan kesadaran pelaku bisnis di Indonesia terhadap potensi data keuangan yang tidak terstruktur dan belum dimanfaatkan secara optimal. Misi ini coba diakomodir melalui upaya untuk menggali nilai yang tersembunyi dalam data keuangan lewat Optical Character Recognition (OCR) dan visualisasi analitik. Fintelite menawarkan solusi untuk memperkaya data keuangan dan OCR yang berfungsi untuk mengotomatisasi proses ekstraksi dan pembersihan data.
Category
📺
TVTranscript
00:00Thank you for joining us on this episode of KreativeKali.
00:07This time, we're going to talk about Cuan,
00:09a financial data-based start-up.
00:11It has been connected through a video conference
00:13with Ms. Nadia Amalia,
00:15CEO and co-founder of Vinterlight.
00:17How are you, Ms. Nadia?
00:19I'm good, how are you?
00:21I'm good. Thank you for joining us on KreativeKali.
00:24Stay healthy for all of us, Ms. Nadia.
00:28Okay, let's talk about Vinterlight.
00:31How can Vinterlight be a solution
00:33for migration of document records
00:35from conventional to digital?
00:38And what efforts can be done
00:41when the data can be extracted
00:44to the digital platform?
00:46Ms. Nadia.
00:47Yes.
00:48So, from Vinterlight itself,
00:50we automate the process of analyzing financial documents.
00:54Usually, if it's done manually,
00:57what happens is we digitize the document.
01:01For example, through photo or scan,
01:03then we input it manually,
01:05and we analyze it manually.
01:07And that process takes time,
01:09about hours.
01:11Many documents that we analyze,
01:13up to 1,000 pages,
01:15it takes one day.
01:17But with Vinterlight technology,
01:19with the two technologies that we have,
01:21from the user itself,
01:23just upload the document or take a photo of the document,
01:26Vinterlight will automatically digitize,
01:29extract, and analyze.
01:31So, the process that took time,
01:33up to hours, 7 hours,
01:35we can shorten that process
01:38to only 45 minutes.
01:40So, on average, we see it's five times faster
01:42than the manual process
01:44with full input manually.
01:47Okay. Because what we have in our field,
01:49when it comes to digitizing data
01:52or maybe financial documents,
01:55like invoices and so on,
01:57it's just a scan.
02:00Then to process it,
02:03we have to input it manually.
02:06We have to type it one by one and so on.
02:08Meanwhile, with Vinterlight itself,
02:10it just needs to be uploaded.
02:12That's about it. Then it's extracted,
02:14and it can also be directly analyzed.
02:16Right.
02:18Up to analysis with AI technology.
02:20Okay. Speaking of technology,
02:22maybe you can tell us,
02:25how does it work
02:27by using AI technology,
02:30Ms. Nadia?
02:31Yes.
02:32So, we use AI technology
02:35that we built in-house as well.
02:37So, we use two technologies.
02:39The first is OCR technology,
02:41or Optical Character Recognition,
02:43where we extract text from documents.
02:46So, for example,
02:48a document like a bank statement
02:50or an invoice,
02:52we extract the words.
02:54Then with AI technology,
02:56we do classification for those words.
02:59And also do analysis.
03:01So, for example,
03:03we move in the field of insurance.
03:05In insurance, there are usually rules.
03:07For example, reimbursement claim
03:09cannot be more than 5 million or so.
03:11Here, with our AI technology,
03:13not only do we extract data,
03:15but we can also do injection
03:17for those rules.
03:19So, the processes that we sorted manually
03:21can also be automated by us.
03:23Okay.
03:24What is the level of accuracy of extraction,
03:27Ms. Nadia?
03:28Because we know that financial documents
03:33are not all in good condition.
03:35There are some that are already cracked.
03:38Then there are notes, etc.
03:41When it is uploaded,
03:43more or less, to the VendorLive platform,
03:46how accurate will the data be read?
03:49Yes.
03:50So, from accuracy,
03:52we see it from 2 sides.
03:54First, accuracy from extracting data.
03:56So, how much data can be extracted
03:59from our technology.
04:01And the second is accuracy in analysis.
04:04Is the analysis correct?
04:06For example, total invoice,
04:08is the number correct with the document?
04:10So, from those 2 sides.
04:12First, from the extraction side,
04:14on average, we can read more than 98%.
04:17Or even for some documents that we have trained,
04:20it is more than 99% to 100%
04:23for the text that can be taken.
04:26The second is from the accuracy
04:28of how much data we can analyze correctly.
04:32And we can also summarize correctly.
04:34This is already above 90%
04:38for our average accuracy.
04:41And one of the advantages of VendorLive itself
04:44is that we can do training
04:47or do the process
04:49without having to do a long training process
04:51because we already have a technology,
04:53a model that is ready to use.
04:55Okay.
04:56If there is a data that is extracted
04:58and there is a lack of space,
05:00a correction that can be done,
05:02what is it like for us?
05:04Is it easy to revise the numbers
05:06that are not correctly extracted?
05:09Because there is still a margin error, right?
05:13What is it like?
05:14That's right.
05:15Well, this is just like humans.
05:17So, for example, humans, on average,
05:19we have a margin of error of 20% to 10%.
05:22Because of that, we believe that the combination
05:25between technology and humans is very important.
05:29So, on our platform,
05:31for example, we are accurate about 90%,
05:34where does the 10% go?
05:36That's for the final review.
05:37That's why on our dashboard,
05:39there is something called Human in the Loop,
05:41where the results of our extraction or digitization
05:44can be reviewed again by humans.
05:47So make sure 100% of the data is correct.
05:50Okay.
05:51You still need skills to check everything,
05:53but at least in the extraction process,
05:56or maybe moving to this digital platform
05:59can be done easily and quickly.
06:02This cuts our work time,
06:04so it's more efficient.
06:05Okay.
06:06What kind of documents can be digitized
06:09or maybe extracted through the Vinterlite platform?
06:11We will discuss the following issues,
06:13Ms. Nadia and Mr. Prisad,
06:14stay with us, we will be right back.
06:17Thank you for joining us,
06:18and I'm still talking with Ms. Nadia Amalia,
06:20CEO and Co-Founder of Vinterlite.
06:23Then, what documents can be extracted
06:26well on the Vinterlite platform, Ms. Nadia?
06:31Yes, in the beginning,
06:33we started with financial documents,
06:36such as invoices, account mutations,
06:39and also the data needed for loan onboarding, usually.
06:44However, along the way of business,
06:47we also added several other documents,
06:49and one of the advantages of Vinterlite
06:51is that we can automate to any kind of documents.
06:55So any document can be automated
06:58through our dashboard no-code.
07:00So for example, there are even our clients
07:02who want to automate legal contracts and so on,
07:06unstructured documents like that,
07:08we can also do it through our platform,
07:11which is the dashboard no-code,
07:13where they can directly automate
07:15without the need for coding or training.
07:17Okay, then what is the system like
07:19from the subscription, Ms. Nadia?
07:23Our subscription is very varied,
07:25because our clients are from SMEs to enterprises.
07:30For SMEs, we have a subscription model,
07:34and also a model where we process per document,
07:37so per volume.
07:39For enterprises, we also have a license,
07:42so a license fee.
07:43So instead of a subscription,
07:46we offer a license where there is no maximum number of pages
07:50that they can process with our platform.
07:53Okay, then how much is the cost
07:55for a Vinterlite client?
07:58It's very varied.
08:03Yes, it's very varied.
08:05We can start from 2 to 3 million,
08:08up to, for enterprises,
08:10maybe the ticket size is much higher than that,
08:13depending on the capacity of the documents
08:16that they want to automate.
08:18Okay, from the level of security itself,
08:21it means there is a kind of data that is...
08:25I don't understand either,
08:27is this data that has been extracted
08:30then collected in Vinterlite's data bank,
08:34or what is it like?
08:36And from the level of security,
08:37what is it like from the data that has been collected?
08:42This is also one of the things that we focus on,
08:45especially from the privacy of the data.
08:48Because as a financial-based startup,
08:51data is very important.
08:53Therefore, we ourselves have ISO 27001 license,
08:58and we also have a license from Cominfo,
09:01and also internally, from Vinterlite,
09:04we support several data deployments.
09:07For example, for our enterprise clients,
09:09the data has never left their system.
09:13So we put our system into their data bank,
09:18so that the data has never left.
09:20This is also one of the things that we actively offer to our clients,
09:25that we have that option,
09:27so that their data privacy is very safe.
09:30And also for SMEs,
09:31for the cloud option, we can say,
09:35where the data bank enters our place,
09:37we have one of the features for data retention,
09:41how long we keep the data before we destroy it.
09:44Like that.
09:45So for data, we really pay attention to it.
09:48Okay, you really pay attention to it and keep it safe.
09:51How many clients do you have so far,
09:53and have you profited so far,
09:54or is there a target for profitability, Nadia?
09:58Yes, for our clients,
10:00we have more than 20 clients for now.
10:04And also for our clients,
10:06it's mostly in the enterprise field for now.
10:10And we are now heading to profit,
10:13because this year we are also expanding
10:16to bring Indonesian-based AI products abroad.
10:20There are several markets that we are now entering,
10:23from Singapore, Malaysia, to the Middle East as well.
10:27Okay.
10:28Is there a similar startup there,
10:30or is it just one in the countries you mentioned earlier?
10:36There are several other options,
10:38especially for startups,
10:40there are about 2-3 options for those countries.
10:44Okay.
10:45What is the target, short-term or long-term,
10:47for Ventalight?
10:50We have two targets,
10:51that's why I mentioned earlier that
10:53one of them is that we are heading to profit,
10:56and the reason is because
10:59we really want to expand to other countries,
11:03such as Southeast Asia and Middle East,
11:06and also this year to America at the end of this year.
11:11The second one is to develop more advanced technology,
11:14especially from Ventalight,
11:15because it also accommodates the needs from abroad.
11:18We will develop several products,
11:21especially in finance,
11:23to accommodate the needs of our clients abroad and in Indonesia.
11:27Okay.
11:28I hope the various targets that Ventalight has
11:31can be implemented well,
11:32and can also provide many benefits to the business world,
11:36and also for clients.
11:37Ms. Nadia, thank you for joining us.
11:39Good luck to you.
11:40Goodbye.
11:41Good luck. Thank you.
11:42Okay.
11:43And before we end this webcast,
11:46we will watch again the movement of ISG.
11:49At 9.28 p.m. WEST INDIAN TIME,
11:52ISG is moving steadily,
11:540.47% at level 7684.926.
11:58And the sectoral movement so far,
12:00for the technology sector,
12:02still leads by 1%.
12:05Top Gainers are Pandi, BRMS,
12:07Brand, Bristoba, SMG, RSI,
12:09United Tractors, and also BRPT.
12:11And the top losers are HUMI, SMEL,
12:13Film, Unilever, Mari, Auto,
12:15Acra, then there is also ABMM, and also BSML.
12:26It's been 90 minutes,
12:27we accompanied you in Power Breakfast.
12:29Hopefully today's discussion can be a fantasy
12:31and a source of information for you.
12:33Stay updated with your information,
12:35only on IDX Channel,
12:36The Transparency and Comprehensive Investment Reference.
12:38And don't forget to watch the Market Review program,
12:40which will air at 10 p.m. WEST INDIAN TIME.
12:43And because the future must be ahead,
12:45I am Investor Saham, I am Wiki Ardian,
12:47I want to share the story of Sembunatun
12:49and also friends who are self-employed.
12:51Thank you. See you.
13:03The Transparency and Comprehensive Investment Reference
13:05is a project supported by the U.S. Department of State,
13:07the U.S. Embassy in the Philippines,
13:09and the Federal Reserve.
13:11U.S. Embassy in the Philippines
13:13Federal Reserve
13:15Federal Reserve
13:17Federal Reserve
13:19Federal Reserve
13:21Federal Reserve
13:23Federal Reserve
13:25Federal Reserve
13:27Federal Reserve
13:29Federal Reserve
13:31Federal Reserve
13:33Federal Reserve
13:35Federal Reserve
13:37Federal Reserve
13:39Federal Reserve
13:41Federal Reserve
13:43Federal Reserve
13:45Federal Reserve
13:47Federal Reserve
13:49Federal Reserve
13:51Federal Reserve
13:53Federal Reserve
13:55Federal Reserve
13:57Federal Reserve