Solving the Blank Canvas Problem: Gusto's AI Co-Founder

Predictions and Expectations:

• Eddie Kim expects that small business owners will use the time freed by automation to focus on business growth strategies, such as expanding product lines and developing long-term strategy.
• Eddie Kim expects that Gusto Co-founder will evolve from a simple business process automator into a proactive partner that alerts customers to compliance issues and eligible tax credits they were unaware of, citing a past instance where the team identified $50,000 in R&D credits for a client named Cabana Pools.
• Eddie Kim expects that the product will enable small business owners to access market intelligence previously available only to the top 1% of technically sophisticated sellers, such as generating weekly competitor analysis reports and pricing optimization suggestions.
• Eddie Kim believes that the "agentic" AI capabilities promised to the world have not yet materialized for the majority of users, who still use AI primarily as a "glorified search engine," and expects Gusto Co-founder to bridge this gap.
• Eddie Kim predicts that the combination of AI and Gusto's existing data will allow the platform to automate complex workflows (like payroll and time tracking) end-to-end without requiring users to log into the Gusto interface.
• Eddie Kim expects that the new development methodology, which relied on five "AI builders" and avoided traditional documentation, represents a significant step change in product velocity that would not have been possible previously.

Timelines and Milestones:

• Gusto Co-founder is expected to add more communication channels, specifically Telegram and WhatsApp, to the roadmap due to the limitations of SMS character counts.
• The product team plans to expand "connectors" to include hundreds of vertical-specific integrations, citing a potential partnership with dental practice management software like Curve Dental as a near-term possibility.
• Eddie Kim expects to open Gusto Co-founder to users who do not yet have a registered company (no EIN) in the "next version" of the product to support side hustles.
• The team expects to assist new users in the next version by automating the process of obtaining an EIN or registering as an employer in California once the user's business scales.
• Gusto Co-founder currently runs on a "heartbeat" trigger mechanism that executes an LLM every 30 minutes, though the system is being improved to support regular cron schedules for deterministic tasks.

Technology and Product Direction:

• Eddie Kim expects the product to move beyond a simple chat interface to include "agentic" behavior where the AI proactively executes tasks based on user intent and historical data.
• The product direction is shifting from a prototype that builds standalone CRUD web apps to one that builds automated workflows leveraging existing customer data and industry-specific context (e.g., dentist verticals).
• Eddie Kim expects that the interface will evolve to support higher bandwidth interactions, potentially moving away from text-only prompts to more integrated data visualization and decision-making tools, though currently focused on SMS and Slack.
• The development process is expected to continue relying on "vibe coding" and rapid prototyping where teams write code to test concepts immediately rather than producing pre-launch specifications or designs.

Market and Industry Outlook:

• Eddie Kim predicts that AI will create a "step function" increase in the ease of starting a small business, leading to a continued secular trend of more business applications being filed with the government.
• Eddie Kim expects that the availability of AI tools will democratize market intelligence and competitive analysis for small businesses, allowing them to compete more effectively with larger entities.
• Eddie Kim believes that small business owners, unlike enterprise clients, will view AI automation as an unqualified upside ("a hard yes") because they lack the resources for the "work before the work" involved in current manual processes.
• The speaker expects the total number of goods and services available to consumers to increase significantly as the barrier to entry for starting businesses lowers.

Company Plans:

• Gusto plans to observe customer usage patterns over time to guide the evolution of the roadmap rather than predetermining specific features.
• The company plans to expand the "co-founder" concept to include proactive business advisory services, such as identifying and filing for tax credits automatically after user permission is granted.
• Gusto plans to add "connectors" for hundreds of vertical-specific software tools to allow data to flow in and out of the automation ecosystem.
• The company plans to continue refining the balance between "heartbeat" based AI processing and deterministic cron jobs to optimize costs and reliability for specific business processes.
• Gusto plans to release a version of Co-founder that does not require an existing EIN or company registration to serve entrepreneurs in the pre-launch phase of their ventures.

Financial Guidance:

• No specific financial figures, revenue targets, or growth percentages were stated by the speakers.

Risks and Caveats:

• Eddie Kim notes that the "blank canvas problem" remains a significant challenge in AI design, where users struggle to utilize powerful capabilities without pre-structured guidance.
• Eddie Kim highlights that "heartbeat" based AI execution is not deterministic, meaning it may not fire at exact desired times, which is a risk for time-sensitive tasks like payroll.
• The speaker notes that while AI makes code generation cheap, there is a risk of building too many features simultaneously without discipline, which could overwhelm customers.
• There is a caveat regarding the cost of running frequent LLM prompts (every 30 minutes), which has led the team to implement logic to detect when a task is better suited for a scheduled cron job.
• The speaker acknowledges that while AI can automate many tasks, it currently lacks the AGI capability to replace high-bandwidth human customer conversations and user research entirely.

Confidence and Disagreement:

• Eddie Kim expresses high confidence that the "agentic" promises of AI can be realized for small businesses by leveraging their existing data and recurring workflows.
• Eddie Kim expresses confidence that the "vibe coding" and minimal documentation approach is viable for zero-to-one product development, stating it allowed them to launch a Tier 1 product in 10 weeks with a team of five.
• Eddie Kim expresses a belief that the interface of AI should prioritize simple communication channels (SMS, Telegram, WhatsApp) over complex UIs, citing his own experience with Open Claw and personal setup.
• Eddie Kim expresses caution against building too many features just because AI makes it cheap, asserting that discipline in selecting functionality is more important than ever.