Show HN: Datachef - full-stack data platform | Hacker News Reader

Show HN: Datachef - full-stack data platform

datachef.com

1 pointjoefusaro872 comments

Hi HN, I’m Joe. I’m a data engineer, and I built Datachef because I got tired of maintaining the disjointed data stack (Fivetran, Airbyte, Airflow, Snowflake, Redshift, S3, dbt, Looker, Tableau, a reverse-ETL tool, etc.) and being a middleman for business users that just wanted to self serve.

Datachef is one managed platform where you can:

- Connect your business systems (Salesforce, Marketo, Hubspot, etc.) and/or databases (Postgres, MySQL) in a few clicks. Your data lake is synced in minutes. - Explore your data via natural language or write SQL directly. - Model how data is combined, filtered and aggregated. Create a shared system of intelligence. - Share models across the org so every query, report and slide deck tells the same story. - Integrate data back into Salesforce, Marketo and HubSpot — lead scores, segments and events, in the systems your team already works in.

We shipped a Claude connector (https://www.datachef.com/blog/datachef-claude-connector). From Claude you can run complex, multi-step/multi-dataset analysis, and create or update your models and sync settings.

I’d love feedback, especially from anyone who has run the modern data stack and has opinions on what consolidating it should and shouldn’t give up.

Open articleView on HN

Politely, “$29,988” and its not clear what the value add is? Not poking in you the eye, maybe prompting you to showcase the product more.

This is kinda in my wheel house as both practitioner and buyer (CFO and now looking to OSS contribute to Ossie after many years of dbt cool aid).

Your prompt says you can connect. Cool, table stakes then what? Explore/model/your other claims, can you deliver that? A semantic / ontological layer you just do that for me after I click connect? What if I have a partial Fabric implementation do your port it over?

If yes, (sorry this tone is a bit harsh but trying to be helpful), how exactly do you understand my business to define these metrics?

On top for $20k, when I can reverse engineer your cost base (my cruelest analysis you YOLO into some SQL over parquet and cheap storage back there, how would I know?) what am I getting? Alternatively and this might be easier, dont justify the $20k, what spend does the $20k avoid! As CFO that would get interest. Do you replace a big league $150k stack of fancy vendors where their enterprise customer acquisition cost is your entire license and bc you can (it seems? Taking you at your word/marketing) basically self serve, youre wedging yourself in into this awesome cost driven product advantage. Now thats mega.

Not sure it that helps but you wanted an opinionated view, def provided you that. Good luck! :)

These are great questions / feedback, thank you.

Answering your last question first:

> Do you replace a big league $150k stack of fancy vendors

Yes, exactly.

You get full and scheduled incremental ETL from source systems, a data lake, an integrated agent, a data modeling layer that the agent can write to and read from, visualization tools (simple charts) and the ability to share your models with colleagues, agents, and BI tools.

Also IMHO your $150k estimate is low and does not factor the personnel cost to deploy and maintain; a seasoned data engineer is >$150k.

> A semantic / ontological layer you just do that for me after I click connect?

We can automatically profile and add context to your data, but the most powerful modeling comes from you giving instructions, just like you would to a business analyst or analytics engineer.

E.g. "Analyze the potential and actual ROI on paid advertising over the past 12 months."

Based on the systems you’ve connected (e.g. Google Ads, Salesforce, etc.) we create a model that gets the spend, the pipeline and revenue generated from ads, inferring pipeline attributable to ads (e.g. lead source or a multi-touch model)

If certain tables or columns are missing from our sync process, we seek approval to add them.

Then you can save this model and make it accessible everywhere - to your colleagues, the Datachef agent, to Claude, to your visualization tools (e.g. Tableau Desktop - Free Edition) wherever you need it. This part replaces dbt.

I’d love to get your feedback on our product, we offer a free trial but mention Hackernews and I’ll extend that to as long as you need to thoroughly vet: https://www.datachef.com

Reply on news.ycombinator.com