In the following exchange, Question Base’s co-founder and COO, Yana Vlatchkova (@Yana), writes about how a meticulous inquiry into customer problems preceded their AI MVP, what the recent, widespread AI push has meant, and where true defensibility comes from.
— Sequencing AI just right in the MVP process
— How the AI hype has aided adoption (and the customer-first next steps)
— Weaving in defensibility when the big players have most structural gains
— Doubling down on a singular advantage all founders share
Sequencing AI just right in the MVP process
My co-founders and I have been on a mission to “Enhance the work of teams” for the past decade. Our previous startup, Swipes, had 1 million users, and our current one, Question Base, has had 200+ teams using us since January 23rd.
We focus on understanding what’s broken in current ways of working.
Whether you’re building an AI startup or another solution, it’s important to understand the depth of the problem you’re trying to solve and the psychology of the users. This is the only way to make a solid bet on a market.
With Question Base, we used exactly this method.
For a few years, we worked as an out-of-house product team for different startups and scaleups. We saw how difficult it was to acquire knowledge within an organization.
As consultants, we were locked out of the company’s systems of record, so we always had to ask questions in the chat. Interestingly, all other employees were doing the same, instead of searching for the answer in the wiki or documents.
We saw that the whole IP of a company was getting shared in the unstructured chaos of the chat and was never retrieved again.
So we learned very basic things:
- The file system with its tree structure needs to go - Noone can find anything; Know-how gets lost at each step;
- Chat was a wealth of undiscoverable knowledge
- Wikis only capture top-down know-how
- Peer-to-peer tacit know-how remains stuck in people’s heads
- People want to ask questions because that’s how they learn best. It digests the information at their level of understanding
- Experts hate answering repetitive questions but do it all the time
The MVP we created was based on these learnings and the thesis that the fastest way to unblock someone is to give them an answer to their question right away.
Initially, we built a very manual system; however, with the use of elastic search and tags, we were able to match a searched question to a previously asked one that had already been answered.
We worked with a pilot company to optimize their knowledge sharing process and prepare them for the hyper growth they were experiencing (from 60 to 200 people in 4 months).
As the limitations of the product became more apparent, we began to see the potential to introduce AI into the product and automate some of the friction points.
One big lesson for us was to shift our focus from recording feature requests to mapping out customer problems. For example, users told us they needed more tagging options so they could organize their data more granularly.
We kept asking why they needed that, and eventually figured out that we could achieve their desired outcome - routing the right content to the right people - much more efficiently with AI rather than tags.
This gave us an edge on the market and we ended up building one of the first AI-powered knowledge bases before these were common.
So, returning to your question about building an AI MVP in a B2B SaaS, the number one priority remains understanding the customer’s pain points very well.
AI can help you optimize a workflow, but first you need to understand the manual processes, the friction, and the daily pains that people are experiencing.
How the AI hype has aided adoption (and the customer-first next steps)
We introduced AI to Question Base in mid-2022, which gave us only a slight advantage at that time.
We discussed the AI revolution, AI tools, and their potential to significantly impact people’s workflow, but the audience was unimpressed.
People were still looking for solutions that could solve their problems quickly.
We found it difficult to position Question Base and get companies to commit to piloting it with us, as well as understanding the topic and communicating its immediate value.
However, this changed in January 2023. With the hype surrounding ChatGPT and their aggressive commercialization, people suddenly became eager for anything AI.
We launched Question Base at the beginning of the month, and since then more than 200 teams have implemented the product. This meant that two years of groundwork had paid off, and we could reap the rewards of having an AI-first solution on the market early.
I loved the interview with Alyona (@zelandiya) and she’s making very solid points. Our experience is very similar - you need to have a customer-first approach in entering a market. The AI hype can accelerate your growth, not create it.
Not in a sustainable way at least.
AI is one of the big advantages companies can apply now to mitigate costs.
In the hard economy we’re facing, everyone is looking to make most out of the bucks they are already deploying. If an AI bot can assist employees in finding the right information when they need it, this has clear monetary value to companies.
We just received a referral from one company to another. Our client has referred us to their friend as an “Additional member on your team for a penny.”
We are still in the early days of understanding the ICP and buying cycle in medium size companies and larger enterprises.
We see QB being installed as a Slack bot by a variety of personas. The overall theme is recusing 80-90% of the internal know-how and all the questions people ask about sales, support, and product knowledge.
And Slack is, of course, our GTM just to get a foot in the door of an organization and deliver immediate value. We are launching our own interface with superior browsing and organizing experience as we speak.
Weaving in defensibility when the big players have most structural gains
It’s a fair statement that startups will be killed or disrupted by the big players entering the space with more powerful AI offerings. This is a necessary stage of the evolution of the technology and the market for it.
If you raised money in 2022 for a video call, note-taking or other summarization startups or answering bots, it can be a scary prospect, as the AI tools are constantly changing.
The big players with lots of data, existing distribution networks, and partnership opportunities are looking to gain market share by introducing AI-based solutions, which puts startups out of the game.
However, it is important to focus on customer value and the groundwork:
Is there a problem that needs to be solved?
How much does it hurt?
One strategy that can be used to gain a defensible position is to own your own data. Startups that leverage GPT3 and other models to apply to someone else’s data are in a very precarious situation.
This model may be beneficial in the short term, as it was for us when we used generative AI to analyze Slack history channels. But for companies to remain competitive in the long run, they need to have their own system of record to collect data in and drive value from it.
There are various ways of building a new system of record out of existing data. You need to choose what is your edge and the business case you’re solving.
It is similar to the analytics space where the same user events can be sent to different for different purposes: visualization, analysis, re-targeting etc. With AI we can now achieve that for company documentation.
Some tools will be best at giving summaries, others at giving answers, others at extracting correlations and data etc. We choose to focus on turning company information into Questions & Answers and being the best at scaling employee know-how across the organization.
Doubling down on a singular advantage all founders share
None of us can predict where the market will go.
However, we can see the trends: Slack, Notion, and others have already added AI. Many others in the space will soon follow suit. Searching across tools has been solved. Generative AI can digest all this information and give an answer.
But is it the right answer?
Who takes responsibility for it?
The shortest path to the right answer still lies with human know-how. Now the question is how to capture it, organize it, and surface it when people need it. This creates an opening for a new database that saves that know-how in this format across all the tools.
There’s a fundamental need for a Question Base; this is what we are betting on for now. Until AI evolves to a more mature and intelligent state in some years, we will also need to reinvent ourselves as everyone else.
The value of already being a major player on the market with data and distribution will be tremendous at that moment. That’s the long-term game we’re playing.
I understand founders worrying about their startups getting disrupted. But the truth is that you can copy a feature, you can copy a product, but you can’t copy a founder.
My real advantage is the way I think.
So I choose to double down on building myself and my ideas up.
Founders have a choice today - to be bystanders watching the “AI revolution” train pass by or they can hop on and be part of making the ride happen. The second option will definitely bring you the most learnings, value and fun.
Related reading from the Relay archives:
— Thematic’s co-founder, Alyona Medelyan, on the problem-solution paradox of AI-first startups