Nov 26, 2024, 10:01 AM ETConference Presentation, Other

How To Start A Dev Tools Company | Startup School

Y CombinatorNicolas Dessaigne

Watch on YouTube →

The LLM observability market is projected to contain dozens to hundreds of competitors, necessitating a clear differentiation strategy.
Approximately 50% of Y Combinator companies are expected to pivot from their initial concept, with a high likelihood that 90% of early code will be discarded during rapid iteration.
Founders are advised to delay hiring sales teams until reaching approximately $1 million ARR and to postpone building dedicated support or marketing teams until after the company reaches hundreds of employees or several years of operation.
Non-technical marketing is expected to be ineffective, with founders and engineers required to lead outreach, documentation, and marketing efforts personally for extended periods.
Developers are expected to perform documentation and support tasks as part of their marketing, with support best handled by engineers to satisfy technical users and inform product development.
Open source is anticipated to be a requirement for libraries, frameworks, and any tools handling sensitive data such as CRMs or EHRs, potentially shortening enterprise sales cycles by a year or more.
Monetization strategies are expected to favor hosting services, open-core models (featuring SSO, audit logs, disaster recovery, and SLAs), or usage-based pricing, while relying on support and services is discouraged due to misaligned incentives.
Pricing models are expected to follow a "good, better, best" structure, with a low-cost self-serve tier for individual developers, a mid-tier for engineering managers, and a high-tier for CTOs requiring sales-led enterprise features.
Sales strategies are expected to shift from top-down pitching to bottom-up discovery, targeting existing enterprise usage to drive expansion and hiring technical staff, potentially titled "product specialists," to engage technical buyers.
Companies are expected to launch frequently, including multiple times on Hacker News, and dedicate "launch weeks" to generate community excitement and validate ideas through votes and comments.
Founder outreach is expected to be necessary initially, with messaging requiring high personalization to avoid being ignored by developers who prefer technical buyers over generic marketing.
Prototypes are expected to be shown to users immediately rather than waiting for a perfect MVP, with tech stacks chosen based on team expertise rather than popularity to enable faster iteration.
Dev advocates are expected to be a difficult role to measure and should be hired from within the engineering team or community later in the company's lifecycle, rather than early.
Community contributions to open-source projects are expected to be rare and often require significant management to ensure quality, so companies should not rely on them for critical functionality.
Founders are expected to personally lead sales until they can no longer effectively sell the product, while avoiding over-selling to non-technical buyers who seek turnkey solutions.
Companies are expected to engage with critics and "haters" on platforms like Hacker News to establish trust, rather than attempting to convince the detractors directly.
Traditional CMO backgrounds are expected to result in unhappiness or ineffectiveness, as traditional marketers often struggle to understand the developer persona and the need for technical depth in marketing materials.
Documentation is expected to be treated as a first-class citizen, with features not considered complete until documentation is written by developers to ensure accuracy and utility.
Companies are expected to start with a "quick and dirty" approach, knowing they will need to refactor the valuable 10% of code later, and to avoid over-building before receiving user feedback.
The bottom-up adoption model is expected to be the norm, where individual developers become champions within larger organizations, though founder outreach remains necessary to initiate the process.
Examples of successful growth and monetization strategies are expected to be found in companies like Stripe, PostHog, Supabase, and Algolia.
Runtime ideas (APIs, infrastructure) are expected to have better incentives and critical importance compared to build-time ideas (docs, QA, testing).
Founders are expected to proceed with feature requests if developers state they could build the solution themselves, or pivot if the feedback indicates a lack of market fit.
Companies are expected to launch on Hacker News frequently to maintain growth, with hundreds of votes and comments serving as a signal of a validated idea.
Support interactions are expected to deliver higher satisfaction when engineers fix bugs and deploy production code quickly, such as within 20 minutes, rather than using generic level-one support.
Companies are expected to avoid hiring too early, only proceeding with recruitment when convinced the product provides sufficient value.
Open-source companies are expected to win deals by leaning into technical aspects rather than presenting as turnkey solutions for non-technical buyers.
Founders are expected to iterate on outreach messages and prototypes quickly, using "do things that don't scale" tactics like side-by-side implementation to learn fast.
The "good" tier of a tiered plan is expected to be priced low to solve problems for individual developers via self-serve, while the "best" tier targets CTOs with a focus on security and compliance.
Sales teams are expected to adopt a bottom-up approach focusing on existing enterprise usage before engaging management, and to hire people who understand developers to gain trust.
Companies are expected to find specific communities such as Hacker News, subreddits, and Discords to establish expertise and trust before attempting to sell.
Documentation is expected to be the first place people interact with a product after the homepage, requiring it to be accurate and written by developers.
Support staff are expected to be engineers who speak the developer language to satisfy users and help product developers understand customer needs.
Marketing is expected to be led by founders and developers for a long time, as traditional marketers are expected to be unhappy or ineffective in the DevTools space.
Companies are expected to be careful when relying on community contributions for open-source projects, as the quality may not meet standards and requires additional management.
It is expected that developers may tell founders to build a feature themselves, in which case the founder is expected to proceed with that feedback or move on.
It is expected that developers will not be excited to open messages that look like standard marketing copy, requiring personalized outreach.
Companies are expected to win deals by leaning into the technical aspects of their product rather than presenting as a turnkey solution for non-technical buyers.
Companies are expected to iterate on prototypes quickly, with the assumption that 90 percent of written code will eventually be thrown away.
It is expected that runtime ideas (APIs, infrastructure) are more critical and have better incentives than build-time ideas (docs, QA, testing).
LLM observability is expected to be an "obvious" idea that faces significant competition, requiring a clear differentiation strategy to succeed.
It is expected that a new DevTool company will need to find customers through outreach first, as nobody knows the company at the start.
It is expected that a "good, better, best" pricing model will allow the "good" tier to solve problems for individual developers via self-serve, the "better" tier for engineering managers, and the "best" tier for CTOs via sales-led approaches.
Companies are expected to avoid hiring too early until convinced the product provides value.
Companies are expected to choose tech stacks based on expertise rather than popularity to enable faster iteration.
Companies are expected to demonstrate the product to users immediately with prototypes rather than waiting for a perfect MVP.
It is expected that developers will hate being marketed to by non-technical teams, requiring marketing to be done by those who understand the persona.
Companies are expected to receive contributions from the community for open-source projects, but most are expected not to count on them due to rarity and quality issues.
Companies are expected to start building their product with a "quick and dirty" approach, knowing that they will need to refactor the valuable 10 percent of code later.
It is expected that the "bottom-up adoption model" will be the norm for DevTools, but founder outreach is still expected to be necessary initially.
It is expected that a clear idea of differentiation is required before starting in the LLM observability space due to the high number of existing competitors.
It is expected that companies like Stripe, PostHog, Supabase, and Algolia serve as examples for successful growth, monetization, and marketing strategies.
Companies are expected to use Hacker News for launching, with the expectation that hundreds of votes and comments indicate a validated idea.
It is expected that developers will be more satisfied if their technical colleagues handle support rather than generic level one support staff.
It is expected that documentation should be written by developers to ensure it is useful and accurate.
It is expected that a "good" self-serve option in a tiered plan will be priced low to solve problems for individual developers.
It is expected that the "best" option in a tiered plan will target CTOs with a focus on security, audit logs, disaster recovery, and SLAs.
It is expected that a sales-led approach is necessary for the highest tier of enterprise plans.
It is expected that founders should personally lead sales until they can no longer sell the product effectively themselves.
It is expected that founders should personally lead marketing for a long time, potentially forever.
It is expected that dev advocates are a difficult role to measure and build accountability for, leading to a recommendation to wait before hiring them.
It is expected that companies can shorten their enterprise sales cycle significantly by being open source.
It is expected that companies will eventually pivot their idea if the first one is not successful, citing that this is a huge reality for YC companies.
It is expected that developers will prefer working with open-source tools over proprietary ones.
It is expected that documentation serves as the first place people interact with a product after the homepage.
It is expected that support interactions where developers fix bugs and deploy production code in 20 minutes will blow customers away.
It is expected that traditional marketing backgrounds will not result in happy CMOs for DevTool companies.
It is expected that companies will face "noise" from many competitors in obvious idea spaces like QA tools.
It is expected that companies will need to engage with "haters" on platforms like Hacker News to convince other readers, rather than trying to convince the haters.
It is expected that companies can leverage the "bottom-up" model where individual developers become champions at bigger companies.
It is expected that companies will need to ask co-founders or developer friends if they would be excited to open an outreach message to gauge its effectiveness.
It is expected that companies will learn as they go and eventually pull off an idea even if they don't differentiate themselves right away.
It is expected that companies can start with a "glorified autocomplete" or minimal product and close a contract if it is significantly better than alternatives.
It is expected that companies will eventually expand from a niche to a broader market once real customers love the product.
It is expected that companies will need to be careful not to misunderstand developer feedback, such as when a developer says they could build it themselves.
It is expected that companies will need to avoid over-building before getting user feedback to prevent wasting time.
It is expected that companies will need to personalize outreach messages to avoid being ignored, as developers hate generic marketing messages.
It is expected that companies will need to iterate on their outreach messages until they get the desired response.
It is expected that companies will need to launch many times to get attention, not just once.
It is expected that companies will need to engage with comments on Hacker News, including those from haters, to show class to other readers.
It is expected that companies will need to do things that don't scale, such as helping customers implement the product side-by-side, to learn fast.
It is expected that companies will need to avoid choosing tech stacks just because they are cool, focusing instead on expertise for speed.
It is expected that companies will need to avoid hiring too early until convinced the product provides value.
It is expected that companies will need to avoid over-building before user feedback to prevent wasting time.
It is expected that companies will need to iterate quickly and launch early to learn.
It is expected that companies will need to consider open source as a primary go-to-market strategy.
It is expected that companies will need to let founders sell the product until they can no longer do so effectively.
It is expected that companies will need to let founders lead marketing for a long time.
It is expected that companies will need to avoid over-selling non-technical buyers.
It is expected that companies will need to avoid over-building before user feedback.
It is expected that companies will need to avoid hiring too early.
It is expected that companies will need to avoid choosing tech stacks for coolness.
It is expected that companies will need to avoid misunderstanding developer feedback.
It is expected that companies will need to avoid over-engineering prototypes.
It is expected that companies will need to avoid waiting for a perfect product to show users.
It is expected that companies will need to avoid generic outreach messages.
It is expected that companies will need to avoid not launching early.
It is expected that companies will need to avoid not doing things that don't scale.
It is expected that companies will need to avoid not engaging with haters on Hacker News.
It is expected that companies will need to avoid not launching many times.
It is expected that companies will need to avoid not personalizing outreach.
It is expected that companies will need to avoid not iterating on outreach messages.
It is expected that companies will need to avoid not engaging with comments on Hacker News.
It is expected that companies will need to avoid not using developers for support.
It is expected that companies will need to avoid not writing documentation by developers.
It is expected that companies will need to avoid not making documentation a first-class citizen.
It is expected that companies will need to avoid not focusing on technical buyers.
It is expected that companies will need to avoid not focusing on existing usage in enterprise accounts.
It is expected that companies will need to avoid not having founders lead sales.
It is expected that companies will need to avoid not having founders lead marketing.
It is expected that companies will need to avoid not having developers on the marketing team.
It is expected that companies will need to avoid not waiting to hire dev advocates.
It is expected that companies will need to avoid not hiring dev advocates from within.
It is expected that companies will need to avoid not being careful about community contributions.
It is expected that companies will need to avoid not starting immediately.
It is expected that companies will need to avoid not iterating quickly.
It is expected that companies will need to avoid not launching early.
It is expected that companies will need to avoid not considering open source.
It is expected that companies will need to avoid not letting founders sell.
It is expected that companies will need to avoid not letting founders lead marketing.
It is expected that companies will need to avoid not focusing on technical aspects.
It is expected that companies will need to avoid not focusing on existing usage.
It is expected that companies will need to avoid not making support a marketing tool.
It is expected that companies will need to avoid not making documentation a marketing tool.
It is expected that companies will need to avoid not using developers for marketing.
It is expected that companies will need to avoid not launching often.
It is expected that companies will need to avoid not finding communities.
It is expected that companies will need to avoid not being helpful in communities.
It is expected that companies will need to avoid not establishing trust.
It is expected that companies will need to avoid not using open source for libraries.
It is expected that companies will need to avoid not using open source for sensitive data.
It is expected that companies will need to avoid not using hosting as a monetization model.
It is expected that companies will need to avoid not using open-core as a monetization model.
It is expected that companies will need to avoid using support and services as a monetization model.
It is expected that companies will need to avoid not using usage-based pricing for APIs.
It is expected that companies will need to avoid not using tiered plans for applications.
It is expected that companies will need to avoid not hiring technical salespeople.
It is expected that companies will need to avoid not using product specialists as sales titles.
It is expected that companies will need to avoid not engaging with haters.
It is expected that companies will need to avoid not launching many times.
It is expected that companies will need to avoid not doing things that don't scale.
It is expected that companies will need to avoid not choosing tech stacks based on expertise.
It is expected that companies will need to avoid not getting user feedback quickly.
It is expected that companies will need to avoid not being careful with feedback.
It is expected that companies will need to avoid not hiring too early.
It is expected that companies will need to avoid not iterating on prototypes.
It is expected that companies will need to avoid not showing prototypes to users.
It is expected that companies will need to avoid not launching on Hacker News.
It is expected that companies will need to avoid not personalizing outreach.
It is expected that companies will need to avoid not iterating on outreach messages.
It is expected that companies will need to avoid not engaging with comments.
It is expected that companies will need to avoid not doing things that don't scale.
It is expected that companies will need to avoid not choosing tech stacks based on expertise.
It is expected that companies will need to avoid not getting user feedback quickly.
It is expected that companies will need to avoid not being careful with feedback.
It is expected that companies will need to avoid not hiring too early.
It is expected that companies will need to avoid not iterating on prototypes.
It is expected that companies will need to avoid not showing prototypes to users.
It is expected that companies will need to avoid not launching on Hacker News.
It is expected that companies will need to avoid not personalizing outreach.
It is expected that companies will need to avoid not iterating on outreach messages.
It is expected that companies will need to avoid not engaging with comments.
It is expected that companies will need to avoid not doing things that don't scale.
It is expected that companies will need to avoid not choosing tech stacks based on expertise.
It is expected that companies will need to avoid not getting user feedback quickly.
It is expected that companies will need to avoid not being careful with feedback.
It is expected that companies will need to avoid not hiring too early.
It is expected that companies will need to avoid not iterating on prototypes.