OfficeBots: (Half) Year Review

It has not been a full year since I started OfficeBots, though it was just the end of the calendar year.

As we go into a new year, I wanted to share OfficeBots’ progress publicly. 😳

We always hear about startups when they have become a success, or sometimes when they have failed (and the learnings gained), not so much when they struggle.

Yet the struggle is a part of the life of any startup. Obviously the ones who failed, but the successful ones too.

Some background

Usually the best path to success is to find a pain point, that people or businesses are willing to pay for, and build the solution to solve that pain.

After selling my first start-up, I was in the somewhat luxurious position of having some financial freedom, thus time on my hands.

As I asked myself what I should do next, I decided to focus on what I liked doing the most previously, which is setting up processes and automating them. I know, I’m a nerd! 🤓

I have a lot of experience setting up processes to help startups scale, and learned programming a few years ago. Those two combined made me extremely passionate about helping people to automate their mundane, manual, repetitive tasks and processes.

When loosely discussing the idea of providing automation-as-a-service, to friends and acquaintances, the feedback was positive. Including a couple of use cases shared with me, as potential candidates for automation.

So I decided to move ahead with that (in hindsight, I should have read The Mom Test before!).

The challenge – that I knew about when I started, but underestimated – was that I had not a particular focus in mind.

Automation of back-office processes is quite broad.

But I thought “how hard can it be to find a scalable use case of repetitive manual work, that I can automate?”.

Well, it’s harder than I thought.

I know… it’s ALWAYS harder than one thinks in startups. I should have guessed 😂

And this is the phase I have been in for these last six months – in parallel with doing some consulting mainly around automation.

The search for product-market fit

It’s called searching for (or finding) product-market fit.

It means finding a product (or service) that the market is ready to pay for, that scales, ideally (or obviously) with good unit economics. And where your product and service is so good, that most clients would feel upset if you decided to stop.

So, despite that knowledge, I decided to launch OfficeBots by building a generic automation platform, on the premise that Bots are reachable by email, which could help democratise automation by providing “digital assistants”. No user interface, so nothing to learn. We setup and maintain the Bots.

Here is an overview of how it looks like:

My bet was that once the MVP (Minimum Viable Product) was up and running, I would be able to quickly test various hypothesis, following discussions with people and companies in need of automation. And so I would find (stumble upon?) one use case I can focus on.

So let me share with you the customers, Bots and automations we have built in the last few months.

Automating Pricing Updates


A company operating in the telecommunications industry, doing wholesale. They buy minutes and data from large telecommunications company, and resell to individual businesses at retail rates.


Receiving, almost daily, pricing updates from their different suppliers, in Excel or CSV format. These are files with 20,000 up to 80,000 records. Each defining a price for a specific country & call.

Each time, manual lookups have to be performed, to:

  • identify price variations which will impact sales prices
  • match the new prices to the prices in the client’s system
  • identify best provider for any given country and call type.

Obviously, any lag or error in ingesting those updated prices in their systems and decision-making process is a potential for costly errors.

And with a daily, hour-long, mind-numbing task like that, errors are bound to happen.

Even during the assessment phase, we identified that the Excel formula used for lookups had an issue (a parameter was missing), meaning a lot of prices had been wrongly assigned in the past!


We built an automation that allows for the client to forward the CSV or Excel file(s) to a Bot, who will perform the tasks required, and confirm the task is done within minutes, returning the processed files.

Ultimately, the Bot’s email address can be shared with the suppliers, so that they email their files directly to the Bot, for it to do its work.

Automating Client Onboarding


A small law firm, specialising in Will and Family affairs.


The company gets 8-10 new clients per month. For each, an hour is spent manually filling out all the forms that the client needs to sign. That’s 8-10 potential billable hours, doing manual, repetitive work.


We built an automation where clients input their data in a secure online form (encrypted), and the Bot generates all required documents with the client’s data (in Word and PDF), using templates hosted by the client in a Dropbox folder.

This makes it seamless for the company to update, add or remove templates – right on their computer – to ensure the next client onboarding uses the latest version.

So far, this is the use case where we are seeing interests from others.

Automating Data Ingestion


Operating in the aviation industry, and ingesting a lot of data from various sources to feed their dashboard, intended to help with decision-making.


Some of the most useful data resides in the flight operating systems used by their client. This data was obtained and ingested via manual work, asking clients over email, and receiving Excel or CSV files in various formats. Which then had to be cleaned up and transformed to match their data schema. Every month.


A Bot was built, connected to the APIs of the flight operating systems, and fetching the data daily. Transforming it and uniformising it to match the client’s systems.

No more back and forth with the clients. No more manual work. And data updated daily vs. monthly.

Automating Financial Reconciliation


Operating in the recruiting industry, the company has a web-app (SaaS) that companies can freely register for.


The finance department spends 2-3 hours every month matching manually companies who have signed up on the web, with the list of their existing customers from their CRM.


We developed a custom fuzzy matching algorithm, as the open-source libraries available were not good enough for this purpose. Fuzzy matching means basically that the Bot can match two company names, and indicate its confidence ratio, even if they are not spelled the same way, or include extra bits. For example “officebots” would be matched with close to 100% confidence with “OfficeBots GmbH.”

Client sends 2 Excel spreadsheets to the Bot, one with the up-to-date CRM data, the other one with the new signups, and the Bot returns the signup list, matched with the company in the CRM.

10mns are needed to review potential false positives, instead of the 2-3 hours it took before.

A (complex) startup’s Venn diagram

There were other small automations we started working on, or explored, but ultimately fizzled out for various reasons.

  • Technical: our solution is entirely cloud-based, which brings some inherent limitations, like not being able to access data “behind the firewall”
  • Business: the use case is not scalable and/or the unit economics not good.
  • Knowledge: the solution requires a deep understanding of a particular field, which – while everything can be learned – adds a lot of risks and complexities to building the right solution.
  • Desire: ultimately, I am building (again) a startup because I want to enjoy what I am doing. So whatever I end up focusing on needs to be something I really want to do (and build) 😇

The Venn Diagram of finding what to focus on MIGHT be a bit complex though! 😅

OfficeBots Product-Market Fit Venn Diagram

Free Bots

We also built a few free Bots.

Some were built based on our own needs, some are simply a single function we extracted from one of the custom Bots we built, and made available for free.

This collection of free Bots is intended to become an “Office Toolbox” – providing easy ways to perform annoying small office tasks.

They are also a way for prospects to test how OfficeBots work – digital assistants performing a small task via email.

Here is the list of those currently Live:

  • ExceleeCE: checks validity of email addresses.
  • ExceleeCK: check what your Excel files are hiding.
  • ExceleeCO: compare 2 Excel files. Highlights the difference in each file against the other.
  • ExceleeCS: converts CSV to Excel (XLSX)
  • ExceleeDD: generate dummy data. Choose between People/Companies/Products and number of records you need. All real data.
  • ExceleeHM: applies a colour scale heatmap per row, based on values.
  • ExceleeIR: insert a row after every other row.
  • ExceleeJN: merge records based on common value (similar to SQL JOIN)
  • ExceleeLU: lookup value in one file, append to the other, based on a common value.
  • ExceleeTS: removes trailing spaces
  • ExceleeUH: unhides all hidden rows, columns and sheets.
  • ImageeOP: reduces image file size.
  • PDFeeIM: converts images (JPG/PNG/TIFF) to PDF
  • PDFeeME: merges PDFs into a single one
  • PDFeeOP: reduces PDF file size.
  • PDFeeSP: splits a PDF file into individual PDF pages.

The last 3 in the list are proving to be the most in demand – so far. If you want to use any of them, you can sign up free (no credit card required) here.

I for one use our own free bots almost daily 😁 – for example to optimise the size of the images used in this blog post.

The challenge for 2020

As you can see, the types of automations built so far around the OfficeBots platform, are quite broad.

Our positioning and messaging is too broad too. Not specific enough.

This poses a challenge from a Marketing perspective – how do we communicate, who do we reach out to – but most importantly, it makes it impossible to scale.

We have not found yet product-market fit, is the proper way to put it.

Or “a solution looking for a problem”, is another way to put it.

So 2020 will be about identifying a specific use case, that we can uniquely solve, and focus on it.

Right now, it looks like Client Onboarding might be a good candidate. It will be the one I will be dedicating most of my time as we start the year. I have some ideas where we could add value, blending my experience in Sales, Customer Success and building optimised & scalable internal processes.

Let’s see where this leads us.

If you have any insights or idea, you are welcome to contact me.

Hope your 2020 has a clearer path 🤞🏼😁