The Company

Mimica was established in 2018 as a result of the founders – Tuhin Chakraborty (CEO) and Raphael Holca-Lamarre (CTO) – meeting in a London cohort of the Entrepreneur First program. Graduating with a small pre-seed grant, the company raised an undisclosed seed round through Episode 1 later in 2018. Mimica launched its first product in 2020; it then raised $6 million in a Series A round that closed in November 2021, led by Khosla Ventures (with repeat participation from Entrepreneur First and Episode 1). Product development is undertaken at Mimica’s London headquarters, with go-to-market operations based in the US. The company currently has over $1 million in revenues. All of Mimica’s current customers – which include AT&T and Dell – are US-based.

The Technology

Mimica provides two complementary solutions enabling organizations to improve decisions about deploying automation: Mapper and Miner. These solutions recognize that automation initiatives’ success depends upon first identifying the right processes to automate and ensuring that these processes are correctly modeled. Mimica has positioned itself as an enabler for those who view the deployment of robotic process automation (RPA) as of paramount importance but whose efforts so far have struggled.

The starting point for Mimica, however, was more theoretical and the foundation for their current go-to-market strategy. The company’s first research and development phases focused on the ability to record interactions between human operators and software-driven processes, learn the tasks themselves, and reproduce those tasks as potential automations. The embryonic company was introduced to RPA and the challenges early adopters were discovering at scale: namely, deciding what to automate and how to create defensible process maps to define potential automation candidates.

The first of the company’s products to make it to market – Mimica Mapper – primarily addresses the second half of that challenge. Mapper employs an installed agent application for Microsoft Windows that creates a discrete trail of activities undertaken by the machine’s user, which is then sent to Mimica’s platform for analysis (see Figure 1). This collection is completed overtly; a Mimica dialog box with record and pause buttons is visible on the desktop of machines where the Mapper agent is running so that it is clear whether the collection activity is in progress. It also means that detailed whitelisting of applications is not required, as Mapper can be paused if applications or processes outside the remit are used. This is important, as the tool is ideally used for the explicit recording of a discrete process that is known to exist. As such, it collects data at a role-based level and produces a comprehensive process map.

In addition to this method, Mapper can also use screengrab and computer vision to customize virtualized environments that are obscured from the standard Windows UI/control method collection mechanism. Much as Mimica avoids application whitelisting, it also suggests that it requires no specific customer configuration to turn this collected data into usable process maps within a week of operation. Mapper uses a collection of AI technologies – including natural language processing (NLP) to understand text and machine learning (ML) to categorize the types of actions being performed – to turn the data collection into normalized and recognizable task steps within business processes. All sensitive data is cleansed, removed from that collection, and sent for processing.

Mimica Miner utilizes the same desktop agent to collect a representative sample of data from an individual user’s machine, but rather than mapping a definitive process, the platform attempts to ascertain whether the process is a suitable candidate for automation across a wide range of collected data. As such, it tends to run on a continuous basis – with start/stop controls still available as required – as it is used to identify the discrete processes that are then fed into Mapper. This proactive approach to automation is designed to head off some of the challenges Mimica believes exist between subject matter experts, business analysts, and developers in analyzing suitability. Just as with Mapper, an intelligence-based approach to defining the steps within a process should be more accurate and faster than one derived from human interviews. Even though Miner collects more data, Mimica believes that it can operationalize the process of identifying candidates with greater precision.

With a focus on read/write/action events to reduce the noise generated when recording a large clickstream, Miner looks for structured and repetitive tasks at volume within the organization and uses the ML-generated knowledge base Mimica has developed to classify its discoveries and suggest how suitable a set of tasks might be. These can then be scored on ease of automation and projected time saving for each, clustered together under high-level process headings (e.g., accounts payable, accounts receivable, etc.).

Pragmatically, Mimica is explicitly targeting large-scale RPA clients where it believes it will find opportunities to help cement the success of that technology while helping inform current or future phases of its use. This also explains why the company entered into a partnership with UiPath in September 2022, not long after Miner was released; as a potential presale or discovery phase tool, Miner is specifically designed to ease RPA implementations. As a value-add to assist the targeting of the processes most suited to automation, its use is, again, rather pragmatic. It is likely over time, however, as this market begins to mature, that Mimica’s capabilities will be less closely coupled to current trends in automation and more a part of a broader set of enterprise process tools.