- Some AI applications turn out to be quite unexpected 😉 For example, companies usually conduct user research when launching or adding new features to their products. But in large companies, this process can take up to six months.
- As it turns out, such research can be conducted instantly — if you survey not real users, but AI personas created based on them 😉 The most surprising thing is that their responses will coincide with those of real people!
- Will companies be willing to pay for such instant benefits? Absolutely! And this means that you can sell them a platform for conducting research like this:
Project Essence
Lakmoos claims that “the time has come when companies will be able to conduct instant research on their users.”
And all this is thanks to the fact that companies will not study their users – but AI personalities representing these users. These personalities live on the Lakmoos platform, each such persona is a generalized representative of one of the segments of the company’s target audience. These personas can even be given names and visualized to make it easier to communicate with them.
The “character” of these personas is created on two levels:
The first generalized level is created by the startup itself, collecting data and surveying people to accumulate a dataset about representatives of different segments of various target audiences. The second specific level can be created by the company itself, adding the results of its own surveys of its target audience to the “character” of these personas. Previously, companies had to survey hundreds or thousands of representatives of a segment of their target audience to obtain statistically significant results.
Now they can select the desired audience segment on the platform, using up to a hundred possible criteria – and get one AI persona to ask the necessary questions.
As I mentioned earlier, the startup collects data about the audience from very different sources. For example, they have generalized and anonymized data on the amount of money in bank accounts for people with different socio-demographic characteristics – so you can ask the AI persona “How much money do you have in your account?” and it will truthfully answer.
Thus, AI personas are also good because you can ask them any questions like “do you believe in God”, “how much do you fear taking risks” – and get statistically truthful answers.
Answers to such questions help better understand the character and habits of your target audience – to come up with new product and marketing hypotheses.
In addition to these questions, AI personas can be asked ordinary questions that companies ask their existing and potential users in regular surveys or focus groups.
Unlike ordinary people, AI personas can be asked any number of questions with any frequency, without spending time and money on collecting more respondents for research.
The problem is that a limited budget for research is a problem even for very large companies. Surveys by the startup have shown that marketing departments of such companies believe that to conduct research of normal quality before launching new products, their research budget should be increased on average 12 times!
The platform not only allows getting factual answers like “no, I don’t plan to take out a mortgage soon” – but also to map the emotions arising from such answers.
For example, it may turn out that the lack of plans for a mortgage is weakly associated with optimism about current housing conditions – rather, it causes sadness about the inability to make such plans. And these are completely different reasons to give practically the same answer to a question – and different conclusions should be drawn from this.
The startup claims that the answers created on their platform by AI personas statistically coincide with the answers of real people who are representatives of the same audience segment. This is confirmed by a report from the research company Ipsos.
At the same time, preparation for conducting user research by traditional methods takes 6 months – including agreeing on research goals, budget, list of questions, defining a contractor, and so on.
During this time, with the help of the Lakmoos platform, a bunch of research can be conducted – and manage to outpace competitors by launching a new feature or new product taking into account the results of these studies.
End clients-companies pay a subscription fee for access to the platform. The startup takes a month to create the first version of AI personas corresponding to the client’s target audience, after which updates and improvements to these personas will be made automatically as the startup collects and updates the dataset.
Another type of startup clients is marketing and research agencies, which can use the Lakmoos platform as a “white label” (with their own branding) to conduct research for their clients.
Lakmoos was created in the Czech Republic in the spring of last year, and now the startup has raised its first 300 thousand euros in investments.
What’s Interesting
A week ago, I wrote about Rally, a startup that raised $8.85 million in investments and developed CRM to accelerate and reduce the cost of user research for companies’ products. The fact that this startup received investments confirms that companies are interested in conducting such research and want to do it faster and cheaper.
Today’s Lakmoos solves the same problem but in an unexpected way. They don’t automate the traditional process but offer a completely different approach made possible by the development of AI technologies.
Although this approach may seem unexpected at first glance, it logically follows from other examples.
Earlier this year, I wrote about Hyperbound, which offers B2B companies to create AI personas corresponding to different representatives of their product-buying companies – “rough sales director” or “polite marketing director”. Companies can train their salespeople on these AI personas. Hyperbound graduated from Y Combinator last year with this platform.
From here, it’s just one step to create AI personas representing the target audience for B2C companies – and Lakmoos has taken that step.
In general, there is a trend towards creating AI personas that reflect individuality. Lakmoos and Hyperbound create generalized individualities.
As for the Dopple startup, which I wrote about last fall, it is a marketplace of AI personas reflecting specific individuals: famous personalities, historical figures, game characters, and other real or fictional characters. Each of these AI personas can be interacted with to get advice, support, or just sympathy. And the whole point is that each persona is not a faceless “font of internet wisdom” like ChatGPT, but an individual with their own character and set of strengths and weaknesses. This startup raised $1.88 million in investments.
Where to Go
This is the emerging trend now — creating AI chatbots, AI assistants, AI personas with distinctive personalities. This is precisely the common direction of potential movement.
The applications of such an approach can be very diverse.
The startup Personal AI, which I wrote about at the beginning of last year, created a platform where you can create your own “digital double”. According to the startup, this double should respond to messages in messenger apps on your behalf. But I think a more lucrative application of the same technology lies in the fact that online course instructors can use their doubles to answer the annoying and repetitive questions from their students 😉 This startup raised $13.7 million in investments.
The startup Chirper, which I wrote about last summer, created an equivalent of Twitter — but where instead of real people, only AI personas created by users can communicate with each other. To create such a persona, the user only needs to briefly describe its history and main characteristics — after which the platform itself will create an extended profile for this persona and endow it with the ability to write its own posts and comment on others. This currently looks like a toy without a clear practical application, but this startup has already raised $750,000 in investments.
Recently, quite a few startups have emerged creating digital employees. I also wrote about a couple of them — Artisan AI (my review), which went through Y Combinator and raised $2 million in investments, and 11x (my review).
These startups also create digital employees who, in the course of their work, need to communicate with other people — salespeople, recruiters, and technical support staff. It would be very interesting and useful to endow them with different individualities as well — after all, different people require different approaches to achieve desired results. When hiring such employees, attention is usually paid to their individual character traits — so why not pay attention to the same when “hiring” AI employees? 😉
In what areas and for what purposes does it make sense to create AI chatbots, AI assistants, and AI personas with individuality? How can this individuality help? What data can be used to create such individuality?
About the Company
Lakmoos
Website: lakmoos.com
Last round: €300K, 05.03.2024
Total investments: €300K, rounds: 1