The archive · Developer & Business Tools · Product decision · 2017–2025
Mimica bets click-capture process maps train agentic AI; $26.2M Series B
EF-incubated startup maps real workflows from clicks and keystrokes to train enterprise AI agents; ARR +570% in 18 months, $26.2M Series B.
Mimica
What the business is
Mimica's platform records how employees perform repetitive tasks — clicks, keystrokes, rules and exceptions — and automatically infers workflow diagrams, first to build RPA bots faster, now as the process knowledge used to train and operate agentic AI inside large enterprises.
Starting capital：$26.2M Series B led by Paladin Capital Group, announced 2025-09-23, with continued backing from Khosla Ventures, LGVP and Entrepreneur First; preceded by a $6M Khosla-led Series A (November 2021) and seed from Entrepreneur First and Episode 1.
How it started
Tuhin Chakraborty (CEO; machine learning at LinkedIn and Stanford) and Raphael Holca-Lamarre (CTO; PhD in computational neuroscience and machine learning) co-founded Mimica in 2017 in London with early backing from the Entrepreneur First accelerator and Episode 1. Their premise was that RPA tools were hard to implement because finding automation opportunities and hand-building process maps was torture; Mimica's first product, Mapper, learned those maps automatically from employee behavior.
What happened
Mapper launched in 2020 and Mimica worked with Dell, AT&T, Hexaware, Experis and Ironbridge; in November 2021 it raised a $6M Series A led by Khosla Ventures to build US sales, competing with FortressIQ and Skan in a market TechCrunch sized at $107B by 2027. As agentic AI became the enterprise story, Mimica repositioned from feeding RPA to being the bridge between how work is done today and how it can be done with agentic AI: captured workflows become process maps that train context-aware agents in weeks, without consultants. By September 2025 ARR had grown more than 570% in 18 months with 30+ large-enterprise customers including multiple Fortune 500s across healthcare, logistics, financial services and manufacturing, and Mimica raised a $26.2M Series B led by Paladin Capital Group, now operating from Brooklyn and London.
No ending yet — it is still running.
Background
Mimica is a process-intelligence company founded in London in 2017 by CEO Tuhin Chakraborty, who studied machine learning at Stanford and built models at LinkedIn, and CTO Raphael Holca-Lamarre, whose PhD was in computational neuroscience and machine learning. It records how employees perform repetitive tasks and automatically turns clicks, keystrokes, rules and exceptions into workflow diagrams.
Its first bet was that robotic process automation was painful to adopt because process discovery was manual: Mapper, launched in 2020, learned process maps from employee behavior so companies could build UiPath-style bots faster. Customers included Dell, AT&T, Hexaware, Experis and Ironbridge, and TechCrunch covered the $6M Series A led by Khosla Ventures in November 2021, in an RPA market it sized at $107B by 2027.
When enterprise attention shifted from RPA to agentic AI, Mimica's maps became the playbook for agents: the company argues that generic agents fail because they don't know how work is really done at each company, and that capturing that context takes weeks rather than months without internal analysts or consultants. By September 2025 ARR had grown more than 570% in 18 months, with 30+ large-enterprise customers including multiple Fortune 500s.
The $26.2M Series B led by Paladin Capital Group, with Khosla Ventures, LGVP and Entrepreneur First continuing, funds moving agents from sandbox experiments to production automation. As of 2026-09-02 Mimica is live and scaling from Brooklyn and London, competing with process-mining incumbents like Celonis and a growing field of agent-orchestration tools.
What has to be true
- RPA adoption stalls on manual process discovery — months of analyst work — so automating the mapping step is a wedge with immediate, provable pain relief.
- The same maps built for RPA become the context layer for agentic AI, letting Mimica ride a new technology wave without starting over.
- Capability without context fails in enterprises, where even simple processes differ by company; capturing real clicks and keystrokes grounds agents in how work actually happens.
- More than 570% ARR growth over 18 months and Fortune 500 customers show the capture-first approach converts, not just demos.
What can be applied
If the data to power a new technology wave lives inside customers' workflows, sell the capture layer first: whoever records how work actually happens owns the map that agents or bots need to act.
Aftermath
As of 2026-09-02 Mimica is live and scaling from Brooklyn and London: the sources reviewed show no acquisition or exit and no round beyond the September 2025 Series B. Its stated direction is moving agents from sandbox experiments to high-impact production automation in healthcare, logistics, financial services and manufacturing. Watch items: whether click-and-keystroke process capture stays differentiated as agent-native companies and process-mining giants converge on the same problem, and whether 'weeks, not months' deployment holds at enterprise compliance standards.
Sources
- Mimica, which automates RPA, raises $6M Series A funding led by Khosla Ventures
- Mimica raises $26.2m to make AI agents actually work in the real world
- Founded & Funded: 2025-Q3 Portfolio News
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card