Alembic
Make manual data entry and labelling a thing of the past.
10 followers
Make manual data entry and labelling a thing of the past.
10 followers
Send us the paperwork — contracts, invoices, claims, scanned images. Alembic's AI extracts every field, decides what's clear, and brings you what needs a human call. No templates. No training.
@james_mcarthur1 nice launch congrats🙌love that it flags what needs a human call instead of just guessing confidence scores. Does the human verification happen on our side via a review dashboard, or is that managed by Alembic?
@priya_kushwaha1 Great question! The dashboard has a Standup area where any issues among your documents are rolled up and presented as relevant questions for how to handle these situations. It learns from your answer, and will autonomously handle decisions, data hygiene, and notifying humans when their input is needed.
Hey Product Hunt 👋
Alembic exists to make manual data entry and messy document processing a thing of the past. We built Alembic to turn files, documents, scans, and other unstructured inputs into reliable, structured, auditable data, without forcing teams to constantly babysit an AI system.
Use it through MCP, Slack, API, or the Alembic app. Pricing is transparent and consumption-based: pay for what you use, and no more.
The problem we kept seeing was surprisingly simple.
Traditional automation struggles when formats change. AI makes this easier, but introduces a different problem: when something is unclear, models can make assumptions, quietly get things wrong, or simply fail.
Alembic is designed differently.
When Alembic encounters uncertainty, it tries to resolve the problem itself first. If human input is genuinely required, it raises a targeted question instead of quietly guessing.
Our Stand Up feature collects those questions into a simple daily workflow, so users can resolve exceptions quickly and get on with their day.
For higher-volume workflows, that same system can operate continuously, allowing teams to resolve document validation and extraction issues in real time.
As we built Alembic, something else became increasingly obvious:
Extraction is only half the problem. Once you've turned thousands of messy documents into structured information, you should actually be able to do something useful with it.
Alembic's reasoning layer, DISTLR, can work across large and complex datasets, manipulate and analyze data inside a sandbox, trace answers back to their sources, and help users understand what all of that information actually means.
That could mean:
Turning hundreds of scanned invoices from different suppliers into one consistent dataset.
Processing travel documents for hundreds of people and automatically grouping flights, hotels, dates, travelers, and missing information into individual profiles.
Collecting documents from customers through a simple link, tracking what's still missing, and updating records automatically when new information arrives.
Or taking years of files and asking questions across them without manually opening document after document.
The goal isn't simply "AI that reads PDFs."
It's a system that can reliably turn messy information into something structured, usable, and trustworthy.
I'm James, the founder and builder of Alembic. After more than a decade working across Customer Success, SaaS, and Data Analytics, I kept running into versions of this same problem: customers needed information in one format, but the real world stubbornly insisted on sending it in twenty others.
Eventually I decided to build the thing I wished we'd had. We're now looking for our first group of customers.
The first 15 companies joining our early adopter program will get their first 4 months free. In return, we'd love your feedback, your ugly edge cases, your bizarre documents, and your help deciding where we focus next.
Alembic is also completely self-serve, with a free tier if you'd rather poke around on your own first.
If you've got a workflow involving messy documents, repetitive data entry, document validation, or information that refuses to arrive in a nice clean schema, I'd love to hear about it.
Email info@alembictransform.com or visit alembictransform.com/contact.
We're early, we're shipping quickly, and this is the point where customer input can genuinely shape what Alembic becomes.