News for hackers. Stuff that matters.

parkerjamie1993

Wow this is awesome!!!
This is awesome! Great stuff guys
Jamie here, one of the Founders of Heron Data. After personally-spinning up many hundreds of classification and merchant enrichment models, we are launching a full self-signup and free trial flow. Anyone can now test out our product without a sales call. You get 2 weeks free use of the API and there is no credit card required. The only difference between the trial and production APIs is a slightly lower rate limit for the trial endpoints. Excited to see what you guys build!

What is Heron Data? We are a YC-backed software company who has felt the pain of working with bank data first hand. We have two product lines which tackle the things we thought most sucked when working with bank & card data:

1) Enrichment: add clean merchant names, icons, URLs, MCC codes, and other metadata to card and bank transactions. These transactions can be from a card you issue, a checking account you host, or pulled via an aggregator like Plaid / Yodlee / Finicity / Tink / TrueLayer / etc. PH launch for consumer applications: https://www.producthunt.com/posts/heron-data-2 & SMB applications: https://www.producthunt.com/posts/heron-data

2) Lending: recreate a businesses P&L + financial ratios from their bank data to better understand companies’ financial health. This is great for any Fintech leveraging bank data or issuing cards for SMBs. PH launch here: https://www.producthunt.com/posts/heron-data-smb-insights

Re:Show HN: NLP and NER-powered beautiful bank transaction feed on Thursday October 28, 2021 @01:28PM (#29028620)
As Ahmed said, the model is trained on UK and North American data but feel free to test it out on Indian merchants too. It will match the bigger / more international ones and still clean up the transaction strings.
You get access to our merchant extraction feature which means for a given transaction description, we clean out the aspects unrelated to the merchant (description_clean), attempt to map it to a canonical merchant and, where we can, return a URL, logo / icon, and unique identifier for that merchant. Latency is very low (c200 ms).

There is a rate limit of 1 call per minute. While we provide more enrichments in our paid tiers we hope that smaller Fintechs in the validation stage can leverage this endpoint in production to build merchant based logic and enhance their transaction feed.

Not in the immediate future - it is really fiddly and not a core requirement for most of our customers.

If this is critical to your guys' use case, let me know! Happy to consider bumping it up in priority.

Thanks for the kind words ram13!

Nope - because we have productised all the ML, merchant extraction, and so on you can get started with a single transaction. We scale up with you as you grow and serve customers who see hundreds of thousands / millions of transactions each day.

We built Heron Data because we experienced first hand the massive time sink of shipping internal tooling to clean up transaction data at our first Fintech.

We don’t want other makers who are pulling data from Yodlee / Plaid / etc. or issuing cards & current accounts to have to reinvent the wheel like we did. We now enrich over 1m transactions each day.

Lots of companies come to us at the idea validation stage while they can’t afford to pay anything. In the past we haven’t been able to help.

We are really happy to launch this free tier today so you can benefit from our cutting edge ML & NER at a lightning fast latency regardless of budget.

If you want a higher rate limit or to leverage our full feature set (which includes recurrence, categories, a Lending Dashboard, and much more), email [email protected].

Re:Pro rata is a bad term for founders on Tuesday October 12, 2021 @04:42AM (#28837462)
Interesting piece. Thanks Aaron!