Sanifu Taught a Machine to Read the Order Nobody Wanted to Type

The order comes in on WhatsApp at 11 p.m. - a customer at a wholesale depot outside Nairobi, thumbing out a list of what he needs by morning. Sometimes it is a photo of a page torn from a notebook. Sometimes it is a PDF attached to an email with no subject line. Whatever shape it takes, someone at the manufacturer has to open it, read it, understand it, and type it, line by line, into an ERP system that expects tidy product codes and quantities. That gap - between how orders actually arrive and how software insists on receiving them - is the entire business of Sanifu.

Sanifu is the product of Patika Technology Limited, a company that went through Y Combinator's Summer 2022 batch as the only startup from East, Central and Southern Africa in the cohort. Its founders - Sidney Rema, Phelix Juma and Bernard Momanyi - spent roughly three years building toward a single, unglamorous idea: that reading a messy order and filing it correctly is a job a machine can now do, and should.

01 / The JobWhat Sanifu actually does

In plain terms: Sanifu reads incoming sales orders and enters them into the systems companies already run on. The inputs are deliberately unfussy - email bodies, PDF attachments, images, and WhatsApp messages, including the kind that started life as a photographed, handwritten note. The AI extracts what matters - who is ordering, which products, how many - matches it against the company's ERP catalogue and product codes, and creates the order. It runs around the clock, which matters when your customers place orders after the sales desk has gone home.

The founders frame the problem with a statistic that anyone who has worked a sales desk will recognize: teams spend as much as 80% of the day keying orders in by hand. That is not a rounding error. It is most of the job, and it is the part of the job that produces the errors - wrong quantities, mismatched codes, missed orders - that quietly cost a manufacturer real revenue.

02 / The CustomersWho is paying for this

The early customers are Kenyan manufacturers whose names sit on supermarket shelves. Kenchic, the poultry company, and Capwell Industries, a grain and cereals processor, both run Sanifu live. Around eleven more manufacturers are in pilots. These are not startups experimenting with AI for the press release - they are established firms with high daily order volumes, where shaving minutes off each order compounds into hours and where a mistyped code becomes a delivery problem.

The reported result is consistent: order-processing time falling from five to ten minutes down to under thirty seconds. The more interesting effect is what happens to the person who used to do the typing. Their day stops being data entry.

03 / The DifferenceWhy not just use OCR?

Plenty of tools can turn an image into text. That is not the hard part. The hard part is the mess: a smudged photo, an order written in shorthand a regular customer expects you to understand, a WhatsApp message that lists products by nickname rather than catalogue code. Generic document-scanning software tends to break exactly where real orders live. Sanifu's work is in the matching - taking an ambiguous line and resolving it to the right SKU in a specific company's ERP - and in being built around the channels African buyers actually use, WhatsApp foremost among them.

That focus is also the honest answer to "who are the competitors." The largest incumbent is not a software company at all; it is the clerk and the spreadsheet. Beyond that sit horizontal OCR and document-processing vendors and workflow-automation platforms - none of them tuned, out of the box, to a Nairobi depot's midnight WhatsApp order.

04 / The ProductsFrom one order to every spreadsheet

Sanifu started at the order and has been moving outward. The flagship remains Sanifu.ai, the order-processing engine for manufacturers and distributors. Over 2025 the company broadened the pitch: automating the wider category of repetitive work that finance, operations, supply-chain and revenue teams currently do in spreadsheets - cleaning data, validating records, matching entries, generating reports. The framing on its own site is blunt about the target: "automate the repetitive work spreadsheets are used for," with integrations, forms, scheduling, dashboards, approvals and AI-powered entity matching stitched around it.

There is a longer history underneath. The company's first product, Patika, helped African small businesses track customer debt and manage cash flow, with users reporting a roughly 17% lift in cash flow and up to three hours a day saved on reconciliations. The move from shopkeeper receivables up the chain to manufacturer order processing is a pivot, but a coherent one - the same instinct to remove manual financial and operational drudgery, applied to a bigger, thornier customer.

05 / The ModelHow the business works

Sanifu sells business-to-business software, priced against a cost a manufacturer can already see on its own payroll and in its own error reports. It plugs into the ERP, email and WhatsApp a company already uses rather than asking anyone to switch systems, which lowers the barrier to a first pilot. The pitch to a factory is arithmetic: here is what clerical order entry costs you, here is what the errors cost you, and here is a tool that takes both close to zero.

The scale of the opportunity is what drew investors. Sanifu points at roughly 180,000 manufacturers and distributors across Africa still processing orders by hand - a market it sizes north of $2.5 billion. It has raised about $1.02 million to date, with Goodwater Capital and Y Combinator among the backers.

06 / The FitWhere it sits in the market

Global enterprise software has spent a decade promising to automate the back office and mostly delivering tools that still require a person to feed them clean inputs. Sanifu's bet is narrower and, for that reason, more testable: start where the inputs are messiest and the pain is most obvious - the order - and prove it in a market that the big vendors have largely ignored. If it works across Kenya's manufacturers, the same problem exists in Lagos, Accra, Dar es Salaam and Kampala, in roughly the same form.

Whether Sanifu becomes the default way African manufacturers take orders, or a feature that a larger ERP eventually absorbs, is the open question. What is not in question is that the problem is real, the customers are paying, and the clock on a single order has already gone from ten minutes to thirty seconds.

ExploreLinks, profiles & a demo

In the press: TechMoran, Tuko and Viral Tea covered the launch and the WhatsApp-to-ERP pitch.

Reporting drawn from public sources including Y Combinator, TechMoran, Tuko and the company's own sites. Figures such as customer counts and funding are approximate and reflect the most recent public reporting.