Mobius makes messy data work.

In 2017, the University of Edinburgh published a contract award for a job that sounds almost offensively old-fashioned: reading handwriting. The material was Scotland’s historical birth, marriage and death records. The winning supplier was Mobius Knowledge Services. The contract value was £1.504 million, excluding VAT. And the procurement contained a pointed instruction: optical character recognition would be unsuitable.

That instruction is a useful entrance into Mobius 365 Knowledge Services and the wider Mobius business. In a market crowded with promises to automate everything, here was a customer explicitly buying human transcription. The interesting question was how to organize it well enough to make millions of handwritten records useful.

The handwriting problem

The Scottish tender required between 18.7 million and 25.2 million records to be transcribed, with setup and transcription completed within 35 months. Sensitive information could be handled only by the successful bidder’s direct employees. Quality carried 60% of the award weighting; price carried 40%. This was a purchase of disciplined execution, with constraints attached. For researchers, the purpose was access to usable records without repeatedly interpreting the original images.

“OCR technology will NOT be suitable for this work”

University of Edinburgh procurement

Mobius describes a 115-person transcription team supported by two project managers. Handwriting varied. Image quality varied. A single tool could not do the whole job. The response included rekeying, spot checks, external validation and a dedicated Process Excellence Group. Its case study reports 99.56% accuracy against a 97% contractual requirement.

The company also reports throughput of 47 records an hour against an expected 39. Those figures describe this particular project. Their broader appeal is the purchasing logic: establish what a correct record looks like, measure the work, and budget for checking it.

A catalog came before the archive

Mobius’s own origin story starts around 2002, normalizing product information for Epinions, the comparison-shopping website. That is a revealing first assignment. Comparison shopping depends on recognizing that differently described items belong together. A bargain is hard to spot when the specifications refuse to line up.

Co-founders Karthik Karunakaran and Gautam Dalmia built a business around that awkwardness. Today, its work includes data extraction, research, integration, digitization, analytics and technology services. The company describes a hybrid human-and-AI approach. Its stated mission is to bridge the gap between available data and information businesses can act on.

Karthik Karunakaran
CEO & co-founder

Gautam Dalmia
Co-founder

The inbox has other ideas

Consider a very different assignment. A UK maritime business received more than 10,000 port-lineup emails daily, carrying attachments in formats including PDF, Excel, HTML and XML. Important details were missing, named inconsistently or buried in incompatible structures. Finding a vessel’s position required more than opening the inbox.

Mobius customized MobiEbot to extract more than 80 port-related data points. The company reports processing about 40 emails a minute. Experts benchmarked quality every fortnight and revised extraction rules using what they found. The reported result was fully automated processing with 85% output accuracy.

Those last two numbers deserve separate attention. Automation measures how much processing happens without intervention. Accuracy measures whether the output is right. For a buyer, the practical implication is to decide which mistakes can be tolerated and which fields need review before anyone acts on them.

Software, with the work attached

The portfolio gives buyers several ways in. PIMworks manages product information and distributes it across ecommerce channels. ProductiWise addresses catalog enrichment. LocationsXYZ supplies points-of-interest data. TechMobius handles technology services. These offerings put Mobius between specialist software suppliers, outsourced operations providers and a customer’s own data team.

The commercial model follows that breadth: projects, continuing services and software platforms. PIMworks says its pricing depends on SKU counts and consumption, with ecommerce services priced to requirements. The Scottish contract offers a concrete project cost; it does not establish a price list for this sprawling portfolio.

In its 2025 retrospective, Mobius identifies the launch of eComNeo as a response to retailers juggling separate catalog, enrichment and price-monitoring tools. The proposed advantage is a shared workflow. That diagnosis makes sense: repairing information in one system is less useful if the next system promptly introduces another mismatch.

The distinction from a standalone extraction tool is the surrounding work. A customer may need source discovery before extraction, classification afterward, and people responsible for exceptions throughout. Mobius offers those activities under the same roof. That breadth is a buying option rather than evidence that every customer needs the entire bundle; a clearly bounded job still needs a clearly bounded contract.

A separate room for automation

There is a distinction to preserve inside the brand family. A September 2024 announcement describes Xtract.io as an independent Mobius spin-off, with a continuing relationship. Its new Chennai office was designed for more than 200 employees. Its flagship Xtract Data Automation Suite connects data sources and automates processing.

Mobius’s customers span information providers, retailers, transport businesses and research organizations. Cengage Learning appears in a named customer testimonial. Its culture pages emphasize direct communication and respectful disagreement. Alongside the automation portfolio, those choices suggest an organization that expects people to remain involved in making the work useful.

The location-data business extends the same reasoning to physical places. In an October 2026 guide, Mobius describes combining points of interest, location attributes and property information for commercial-real-estate decisions. A map pin alone cannot tell a retailer whether a site has the right surrounding businesses. The value comes from connecting attributes to a question someone actually needs answered, then keeping the underlying information current as businesses open, close or move.

Start with the field that matters

For a prospective customer, the copyable approach is modest: choose a representative sample, define the output fields, agree on accuracy and specify delivery. Mobius’s aviation case, for example, describes XML feeds delivered into a client workflow, covering 93,000 flights across 183 airports after a four-week build.

A self-service tool may suit a small, stable dataset. Mobius’s combination becomes more relevant when sources change, formats multiply and someone must keep checking. The archive and the inbox leave buyers with a precise question: who is responsible for the record that comes out wrong?