MILabs
Tools

Marker PDF conversion for business documents | Setup and use

Marker is an open source tool that converts PDFs and other documents into structured text, set up by MILabs on infrastructure the business controls.

Updated 2026-09-11

What Marker is

Marker is an open source tool from Datalab that converts documents, most often PDF files, into structured text. It reads the layout of each page and returns the content as Markdown, with JSON and HTML output also available, so the material can be searched, edited or passed on to other software. The project publishes its source code on GitHub, and it can run on a company's own computer or server as well as through a hosted service. For a small business, the real gain lies in what becomes possible afterwards. A folder of scanned forms, supplier terms or past reports turns into text that a person can search and a language model can read. Teams often start with one folder that is hard to search, then widen the work once the results read well. Marker does not decide what the documents mean. It prepares them, and the business decides how the output is stored, used and put to work.

Datalab: official site

Marker in plain language

A PDF is built to look the same on every screen, not to describe what its contents mean. Behind a page of text there is often a set of characters placed at fixed positions, with little sense of which words form a heading or which belong to a table. People read that layout without effort. A search tool or language model reading the raw file often does not.

Conversion rebuilds that structure. A useful result keeps headings as headings, lists as lists and tables as rows a machine can read. Markdown is a common target because it is plain text with light markup, so it opens in any editor, stores cheaply and moves into other software without special tools.

Clean digital PDFs with consistent layouts usually keep the most detail. Scanned pages rely on text recognition first, so crisp print and straight pages give the richest results. Dense financial tables, mathematical notation and irregular forms add more to work through, and a reviewer who knows the material is the best judge of where the text is strongest.

A converted file becomes useful once someone decides where the text lives, how each piece is labelled with its source, and when it is refreshed after the original changes.

What businesses use it for

  • Converting scanned forms and PDF applications into searchable text for the record.
  • Making policy, contract and supplier documents easy for staff to search.
  • Supplying clean text to a private question and answer assistant built on company files.
  • Lifting headings and body text out of long reports so reviewers can move quickly.
  • Preparing older documents for archiving or for a move to a new system.

Who it suits

  • Law firms and accountants who handle large volumes of client PDFs
  • Property managers and councils reviewing leases and notices in bulk
  • Teams that want document work to stay on hardware they own

How MILabs sets it up

MILabs begins with the owner or team lead, agreeing which document sets matter and where the converted output should land. Because Marker runs on a server the business controls, the first conversation covers where the files may live. We then connect the systems the business already uses, such as a shared drive, a document store or the assistant that will read the output, so the conversion works on the right folders and the results appear where people already look. Permissions are set so that only the people who need the documents can reach the output. We also agree in writing who checks conversions, who updates Marker and who responds when a file fails to convert. The business keeps full ownership of its files throughout.

Builds it fits

Industries it fits

Marker in practice

Inside a business the work usually starts with a shared folder that nobody has time to search. An office manager or compliance lead picks a bounded set of files, such as supplier agreements or one category of scanned forms, and the conversion runs against that set. Converted files sit beside the originals with matching names, so a reviewer can compare them directly.

A reviewer who knows the documents reads a sample and marks where headings, table rows or figures came through cleanly. Those marks show which file types can run with light checking and which benefit from a closer look. Patterns tend to follow layout, so the list of tricky cases becomes a useful record the business keeps.

Once trusted, the text feeds the systems that depend on it. A question and answer assistant reads it when a staff member asks about a policy, and a filing routine matches new scans against the converted index. Each connection needs a named owner who knows what to do when a source document is revised.

Upkeep is a steady routine. When Marker publishes a new version, someone runs it against the sample set first, then it replaces the current one.

good questions, straight answers

Frequently asked questions

How do I get my team to use Marker?
Start with one document set that is slow to search, such as a pile of scanned forms or a folder of supplier terms. Show the converted text beside the original so staff can see how faithfully it reads. Then add the next set once the first is in everyday use. Adoption tends to follow trust in the output, and that trust grows as people check results on work they know well.
Can I self-host Marker?
Yes. Marker's source code is published on GitHub, so it can run on infrastructure the business controls. MILabs can set it up on a machine or server that matches the business's rules about where client files may be processed. The right choice depends on how sensitive the documents are and who will look after the installation. A hosted option is the alternative, and the comparison is worth discussing.
Which builds work well with Marker?
Builds that answer questions from a body of documents suit Marker best, because the conversion step gives the assistant readable text to work from. The document question and answer assistant and the private version that runs on the business's own server are the closest fits. The compliance checklist assistant and the policy compliance assistant also rely on policies held only as PDFs. Invoice capture can use the same step for supplier paperwork.
What is Marker?
Marker is an open source document conversion tool from Datalab. It takes PDF files and other documents and returns their content as Markdown, with JSON and HTML output available. Its code is on GitHub, so a business and its technical advisers can read how it works and run it themselves. It is a conversion step rather than a finished product. Businesses use it to make documents readable by search tools and language models.
How does MILabs set up Marker?
MILabs starts with the owner or team lead, agreeing which documents matter and where the output should go. We install Marker on infrastructure the business controls, connect it to the shared drives and systems in use, and set permissions so only the right people reach the results. Before handover we agree who checks conversions, who keeps Marker updated and who responds when a file fails. The setup leaves ownership of files with the business.
let's make something happen

Workflow automation: how we can help

Choose how you want to work on this with MILabs. Both options start with an enquiry.

Discovery is free, no obligation from the first chat.

Let us build it for you

We build it and look after it.

Build it yourself, we'll help

You build it with our direction, teaching and a weekly check-in.