YC Partners Answer Your Questions | Office Hours

Key Concepts

• Go-to-Market Strategy for AI in Legacy Industries: Approaches for introducing AI solutions into established sectors like accounting.
• Automation Rate: A key metric for companies aiming to automate manual tasks, particularly in service-based industries.
• Founder's Role in Automation: The advantage of software founders in identifying and implementing automation opportunities.
• Scaling vs. Automation: The risk of scaling manual operations before achieving significant automation.
• Technical vs. Non-Technical Staff Ratio: A framework for maintaining a balance to ensure continued development and automation.
• Forcing Functions: Strategies to compel automation, such as limiting hiring.
• Trajectory of Automation Rate: A more critical metric for investors than raw revenue in certain AI startup contexts.
• MVP Development with Industry Partners: Leveraging existing businesses to build and validate Minimum Viable Products.
• Customer Qualification: Identifying early adopters and incentivized decision-makers for new software.
• Learning Pace: The importance of rapid learning from customer feedback for early-stage startups.
• Mid-Market vs. Enterprise Sales: Strategic considerations for choosing initial customer segments.
• AI Sales Software (AI SDRs): The effectiveness of AI-powered sales development representatives when integrated into established sales processes.
• Founder's Role in Sales: The necessity for founders to master sales fundamentals before relying on AI tools.
• Growth Hacking in Startups: The inapplicability of traditional growth hacking techniques from large companies to early-stage startups.
• Model Leap: The impact of advancements in AI models on product relevance and development strategy.
• Pivoting: The process of changing a startup's direction, especially when initial traction is insufficient.
• Conviction: The internal belief and energy required for a successful pivot.
• Valuing the Product: Assessing whether customers genuinely perceive value in a startup's offering.
• Technical Difficulty as an Advantage: How complex technical challenges can create defensibility.
• Scope Reduction: Strategies for breaking down technically challenging projects into manageable parts.
• Hiring Timelines: Determining the optimal time to hire new employees based on workload and breaking points.
• Opportunistic Hires: Hiring individuals with exceptional skills and existing relationships during opportune moments.
• Open Sourcing Enterprise SaaS: The strategic advantages and disadvantages of making enterprise software publicly available.
• Trust and Transparency: How open-sourcing can build confidence in enterprise clients, especially concerning data privacy.

Go-to-Market Strategies for AI in Legacy Industries

When bringing AI to legacy industries, founders face the challenge of delivering a long-term vision of full automation with agents or LLMs, which is not feasible on day one. Three primary go-to-market approaches are discussed, using the accounting industry as an example:

1. Build an AI Software Company Selling to Accountants:

   • Methodology: Understand the accounting industry, identify valuable and achievable AI software development areas (e.g., specific tasks that can be automated in the first 6 months), and build a solution that excels at that specific function.
   • Pros: This is the most common and often successful approach for YC companies. It allows for focused development and selling a valuable, albeit limited, service.
   • Cons: Requires significant understanding of the target industry and the ability to deliver a valuable, purchasable service.

2. Start Your Own Full-Stack Accounting Firm:

   • Methodology: Establish a new accounting firm that handles all aspects of the service, including taxes, closing books, and less common tasks.
   • Pros: Direct control over the entire operation.
   • Cons: Requires significant manual work, potentially needing an accountant on staff. The key metric to track is the percent of work automated, which should increase over time. A failure mode is getting bogged down in manual execution and delaying automation. Software founders are better positioned to identify automation opportunities within this model.

3. Buy an Existing Accounting Firm and Ingest AI:

   • Methodology: Acquire an established firm and integrate AI into its operations.
   • Pros: Immediate customer base.
   • Cons: Significant challenge in changing the culture of an existing company, especially larger ones. This approach is less commonly seen.

Key Argument: For the second approach (starting a new firm), a critical failure mode is scaling revenue too early before sufficient automation is achieved. This can lead to managing a manual firm with some software, which is inefficient. A framework like maintaining a certain percentage of technical staff (e.g., 30%) can help ensure continued focus on automation. Forcing functions, like limiting hiring to one accountant, can also drive automation. Investors may prioritize the trajectory of the automation rate over raw revenue for such companies, as it proves the ability to write software for task automation.

Real-world Application: Vessence, a company building software for lawyers, leveraged a large law firm in Stockholm to work out of their office and build their MVP, demonstrating a successful approach for founders without direct industry exposure.

Customer Qualification: When selling to legacy industries, it's crucial to find customers (firms or decision-makers) who are highly bought into using new software and are incentivized to increase its adoption. This is a stage even before early adopters, requiring pre-qualification to identify individuals who are empowered and excited about new technology.

Mid-Market vs. Enterprise Sales for AI Companies

The discussion shifts to the strategic choice between targeting the mid-market or enterprise for AI companies, considering long sales cycles and investor impatience for growth.

Key Argument: For early-stage companies, the pace of learning is paramount.

• Enterprise Sales: Can lead to slow learning cycles due to long sales processes and delayed feedback. Unless there's a pre-existing "in," it's difficult to learn quickly.
• Mid-Market/Smaller Companies: Offer a faster learning curve, better feedback, and more agility for iteration and product changes.

Caveat: Some problems are inherently enterprise-level, forcing companies to target larger clients from the outset. In such cases, reducing the scope to target a few users within an enterprise or developing a more narrow product can shorten sales cycles.

Qualification is Crucial: Regardless of market segment, qualifying the individual buyer is critical. They must be empowered, incentivized, and the sales interaction needs to be genuine. The right empowered person in a mid-size company can move faster than enterprise sales cycles.

Data/Statistics: No specific figures are provided, but the concept of "multi-million dollar deals" in enterprise and "half million dollar ARR contracts" are mentioned as benchmarks.

AI Sales Software and Founder's Role

The question of whether to hire human sales roles or replicate them with AI is addressed.

Key Argument: AI sales software, particularly AI SDRs, are most effective when integrated into a well-functioning sales process. They are not a solution for founders who are unable to sell their product.

• Founder's Responsibility: The hard work of figuring out how to sell the product remains with the founder. Traditional growth hacking learnings from large consumer companies are often not applicable to startups.
• "Magic Tricks": The two fundamental questions for founders are "Who am I selling to?" and "How do I get their attention?" Once these are figured out, AI SDRs can handle the "shle work" of execution.
• AI SDR Company Pitfall: Companies selling AI SDRs to startups that haven't figured out their sales process often face high churn. The goal should be to sell to companies with a good product that can already sell, enabling them to scale with AI.

Analogy: This is similar to hiring the first salesperson, which is often too early unless the founder has already established a playbook. AI SDRs are an extension of this principle, requiring an established process. Founders should be curious and learn these roles themselves before scaling.

Investing in AI Development vs. Waiting for Model Leaps

Founders are advised on whether to invest aggressively now or wait for future AI model advancements.

Key Argument: Founders should assess if their product will become irrelevant with new models or if it will be enhanced by them.

• If Enhanced by New Models: Investing now, even if some effort is "wasted" before a model leap, leads to significant learning. This learning allows for a much better product on day one when new models are integrated.
• Examples: Cloud Sonnet and Cogen tools saw their functionality dramatically improve with new models, making earlier investments in understanding the space valuable.

Pivoting: When and How

The discussion delves into the complexities of pivoting a startup.

When to Consider Pivoting:

• When things aren't working: This is the most straightforward scenario.
• When traction is insufficient: Even with some revenue and customers, if growth is slow and market needs are evident elsewhere, a pivot might be necessary.

Case Study: Firefly (formerly Mandible):

• Had significant traction ($100k+ ARR) with a Q&A on documentation product.
• Observed slow growth and a market need for a crawler component they built internally for their own product.
• Pivoted to focus on the crawler, which became a more valuable product. This was not an overnight decision but involved experimentation.

Key Argument for Pivoting:

• Customer Value: The core question is whether customers truly value the product. If not, it's a strong signal for a pivot.
• Founder Conviction: Pivoting requires immense energy and conviction. Founders must believe in the new direction, even when starting from scratch.
• Framework for Pivoting: While there's no strict algorithm, deep conviction from customer conversations is key. It's beneficial to explore a range of ideas during a pivot to find conviction.
• Vulnerability: Pivoting is a vulnerable step where companies often fold. Founders need the energy to persevere through uncertainty.

Example: Algolia: Started with an on-device SDK for mobile apps, found sales difficult, and pivoted to a more successful model.

Example: Greyle (Winter '24 Batch): Had a few thousand dollars in MRR but realized through user interviews that customers weren't articulating consistent value. This led them to re-focus and find a better product-market fit.

Leading Indicator for Pivoting: A founder's loss of belief in the current product's success is a strong signal to consider a pivot.

Killing a "Good" Idea for a "Great" One

The distinction between a "good" and "great" startup idea is explored.

Key Argument:

• Greatness is Validated: A great idea is not known in the moment but is revealed through customer feedback and validation.
• "Good" vs. "Great": The speaker posits that ideas are either "great" (yielding a huge company) or "everything else" (technically bad startup ideas).
• Testing for Greatness: Founders must aggressively test their ideas through sales, building wacky versions, and seeking honest feedback to uncover signs of greatness.
• Obsession: Top-performing founders are obsessed with finding a great idea and ensuring they are on that path. They rarely declare their idea "great" prematurely.

Example: Steve Jobs's tendency to call his ideas "dopey" even if he believed in them.

Technical Challenges and Pivoting

The question of whether to pivot away from an idea due to technical difficulty is addressed.

Key Argument: Technical difficulty can be an advantage, creating a higher barrier to entry and less competition.

• If a founder has the courage and skills, a technically challenging idea can be the best one.
• Example: Brahante Biologics, building microfactories for drug manufacturing, faced immense technical and regulatory challenges but had strong founder conviction.

Managing Technical Difficulty:

• Reduce Scope: Break down the problem into smaller, manageable pieces.
  • Example (Perfect Audience): Built a front-end for an existing real-time bidding platform with an API, then eventually built their own infrastructure.
  • Example (Optimizely): Created a janky bookmarklet for manual JavaScript editing to consult on A/B tests before building the full website editor.
• Caution: Founders should avoid using technical difficulty as an excuse to avoid customer interaction. Spending time with customers, even with an incomplete product, is crucial.

Analogy: Founders can become their own users by building a rudimentary version, which helps them understand the problem and validate the need.

Hiring Guidelines for Startups

The optimal time and process for hiring beyond the founding team are discussed.

Key Indicators for Hiring:

• Overwhelm: The right time to hire is when things are so busy that founders cannot find time for interviews. This indicates the company is at a breaking point and needs support.
• Specific Breaking Points: Identify specific areas (engineering, sales, onboarding) that are failing or about to fail.
• Early Warning Signs: Be honest about whether these are genuine early indicators or just hopes.

Hiring Process:

• Start with Personal Network: Early hires are often people who already know and trust the founders.
• Opportunistic Hires: Hiring a "smartest friend" or someone with exceptional, superlative qualities who happens to be available at the right moment.
• Caution: Avoid hiring based on past company prestige alone; focus on genuine talent and fit.

Hiring as a Metric: Hiring is not a success metric itself. It's a tool to enable the company to function and avoid failure.

Phases of Hiring:

1. Pre-Product-Market Fit: Hire only a few key individuals.
2. Post-Product-Market Fit: Hire aggressively to scale, but be mindful of not hiring too late.
3. Over-Hiring: A phase that can occur after significant growth.

Advice: Founders are often advised against hiring too early, as it can slow down progress. The exception is opportunistic hires. Founders should be curious and learn the roles they intend to hire for.

Open Sourcing Enterprise SaaS Products

The advantages and disadvantages of open-sourcing enterprise SaaS products are examined.

Advantages:

• Trust and Transparency: Crucial for enterprise clients, especially those dealing with sensitive data. Open-sourcing allows them to inspect the code, fostering trust and potentially shortening sales cycles.
  • Examples: Medplum (open-source EHR), 20 (open-source CRM).
• Self-Hosting and Compliance: Enables customers to host the product themselves, addressing concerns about data privacy and compliance.
• Extensibility: Customers can expand or customize the product as needed.

Common Use Cases:

• Dev Tools: Developers often prefer open-source products for transparency and control.
• Enterprise SaaS with Sensitive Data: Products in fields like healthcare (EHR) or CRM benefit from the trust generated by open-sourcing.

Drawbacks:

• Cost of Self-Hosting: Customers who self-host incur costs, which needs to be factored into pricing strategies (often requiring a high price for self-hosted options).
• Community Management: While not always the primary goal, managing an open-source community can be resource-intensive.

Trend: The ability for small startups to quickly offer self-hosted solutions is increasing, making it a more common offering. This is a significant shift from the past where self-hosting was often seen as an impossible request.

Conclusion: Open-sourcing can be a powerful strategy for enterprise SaaS, particularly for building trust and addressing data privacy concerns, even if the target audience isn't developers. The ability to offer self-hosting is becoming a key differentiator.