At Lithic, we're building a company of Builders. Not in the Silicon Valley platitude sense, in the literal sense. Everyone builds.

Let me explain what that means, how we got here, and where we think this is going.

Builders and Operators

For most of tech history, companies have been split into two camps: Builders and Operators.

Builders: Product and Engineering — wrote the code, shipped the features, and designed the systems. Operators: Operations, Support, Finance, Marketing, Sales, Legal, HR — made the product actually work for customers. Builders also built the internal tools that Operators used, which meant Operators were perpetually waiting in line for engineering capacity to solve their problems.

This divide was never clean, and the best companies have always tried to blur it. But there was always a ceiling. Operators could push spreadsheets and manual processes only so far. Real software still required real engineers, and that dependency created a permanent bottleneck between the people who understood the problem best and the people who could build the solution.

That ceiling is gone.

The Lithic Builder Vision

Our CEO, Bo Jiang, has pushed a vision for what Lithic should become: a company where everyone is a Builder. Not as a title. Not as an aspiration. As a practice.

Everyone, from Product and Engineering to Data Analysts to Marketing to Risk to Legal to HR, is expected to be able to build tools, automate workflows, and solve their own problems with the help of AI. And they are.

Here's what that looks like in practice:

Our Head of Support built his own customer support dashboard. Not a feature request to engineering, he built it himself. He can view support tickets, investigate issues, and resolve them using a custom tool designed for his specific workflow. No one knows his pain points better than he does. Now no one is better equipped to solve them.

A Data Analyst noticed how much time we were spending manually converting order forms and contracts into line items in our billing and invoicing systems. He built an AI pipeline that translates contracts into billing SKUs automatically. Work that used to consume hours of manual effort, reading contracts, cross-referencing terms, generating billing information, now happens in minutes, if not seconds.

A member of our legal team built an AI integration directly to our CRM, enabling the team to ask natural language questions about existing contracts. Work that previously meant manually searching through and cross referencing documents, now happens in a single query.

Our Product Designer shipped a mobile feature she was passionate about but that never rose to the top of engineering's priority list. So she built it herself, making production changes to our mobile app directly with engineering oversight but without engineering as a bottleneck.

This is what a company of Builders looks like. They saw problems, and they were empowered to solve them.

Building the Company of Builders

These stories didn't happen by accident. Over the past year, one thing we've learned is that AI transformation is not primarily about technology. It's about organizational design. The hardest problems in technology adoption aren't technical, they're human. It's not enough to pick the right tools. You have to transform how people and organizations work.

Governance First

Before a single tool was purchased, we worked with our legal team to establish a clear acceptable use policy for AI. For a company handling sensitive financial data on behalf of clients and their customers, this wasn't a checkbox exercise. We wanted every employee to understand not just what the rules were, but why they existed, so that judgment calls in ambiguous situations would land in the right place.

The most critical requirement was straightforward: any AI tool adopted at Lithic could not train on our data. Full stop. That ruled out a number of otherwise capable tools and became the primary filter in our evaluation. Beyond data protection, we established practical guidance with real examples of acceptable and unacceptable use cases, so engineers could move confidently rather than cautiously. Governance enabled speed. It didn't slow it down.

Let a Thousand Flowers Bloom

With guardrails in place, we gave every engineer a budget to purchase and try whatever AI tools they were curious about, then asked them to share findings through dedicated channels and knowledge-sharing sessions.

This approach was deliberate. Like every major technological shift before it, from  dotcom, to mobile, to cloud, real adoption comes from the grassroots. Picking technologies top-down doesn’t yield the same success. The AI landscape was moving fast enough that no single evaluation process could keep up. What was state-of-the-art one month often wasn't three months later. Rather than lock into long-term vendor commitments, we created conditions for organic discovery and let the tools that actually worked in real workflows prove themselves.

The signal that came back was genuine because the process was genuine. Engineers weren't evaluating tools against a rubric. They were using them to do real work and sharing what helped and what didn't.

Trust Compounds

One of the more important observations from the past year is how AI adoption evolved as people built trust in the tools.

Early on, the use was narrow: AI as a sophisticated autocomplete, useful for syntax, boilerplate, and quick lookups. As trust developed, engineers started delegating more - pair programming, handing off discrete tasks, validating output rather than writing everything from scratch. More recently, the scope has expanded significantly. It's now possible to delegate an entire project workstream to AI, requirements, design, implementation, and testing, with an engineer directing and reviewing rather than writing line by line.

That progression wasn't mandated. It followed naturally as the tools proved themselves and people developed intuition for where AI output was reliable and where it needed closer review. Building that intuition takes time, and it's one reason we didn't try to shortcut the process.

The results were hard to ignore. Engineers using AI saw a 3-5x increase in velocity versus those who didn’t , and the quality gains were just as significant. As one of our Engineering Directors put it: "Rather than shipping v0 or v1 of a product, we are now shipping v4 or v5." The time that once went to first drafts was now going to refinement, edge cases, and polish.

AI With Company-Wide Context 

Last year, the AI landscape was crowded with competing visions. Every vendor had their own idea of what AI adoption should look like, and every one of them wanted to be the center of your workflow. Without a clear point of view, it would have been easy to end up with a dozen disconnected tools and no coherent strategy.

We needed to align on a vision of our own. After evaluating the landscape, we landed on a clear thesis: a single AI platform would become our "window pane" to work, the primary surface through which people across the company access information, build tools, and get things done.

The insight behind this was simple: AI can process and synthesize information across sources far better than a human tabbing between twelve browser windows. Instead of bringing people to the tools, bring the tools to where people already work.

We identified vendors that fit into this paradigm and started building internal tooling to access our systems from within our AI platform. A year later, that vision has proven out. AI is now the kick-off point for most work across the company. Whether you need to search for company information, write documentation, analyze data, or draft a contract review, it starts in one place.

Skating to Where the Puck Will Be

There's a well-worn Gretzky line about skating to where the puck will be, not where it is. In a fast-moving technology landscape, it's worth taking seriously: by the time you finish building for today's limitations, the underlying technology has already caught up with you.

So where is the puck going?

The Builder-Operator Convergence is Accelerating

The line between these roles is blurring fast. Once Operators experience the ability to unblock themselves, to solve their own problems without waiting in an engineering queue, they don't want to go back. And they shouldn't have to. The immediate, tangible problems they face every day are best solved by the people closest to them. As an organization, our job is to enable them to build safely, with the right guardrails, so that this momentum scales without creating risk.

The Arrival of AI Agents

We are just starting to see the next shift: AI agents that don't wait to be asked. Rather than humans commanding AI to accomplish a task, agents will proactively complete work in the background for humans to review. We're already experimenting with agents that monitor and triage support tickets, agents that proactively scan accounts and transactions for compliance, and agents that sit in Slack channels to assist teams, surfacing information and performing actions in real time. This is early, but the trajectory is clear: the Builder's toolkit is about to get significantly more powerful.

AI Getting Smarter with Context

AI will get smarter, not just because of better models, but because it will have greater context on the domain and the company.

Today, AI often fails not because it lacks capability, but because it provides generalized solutions that don't account for the specific needs and context of the organization. Imagine hiring the world's best engineer, product manager, salesperson, or data analyst and dropping them into your company with zero onboarding and zero context. They'd struggle immediately. Raw talent without institutional knowledge doesn't ship.

We're already seeing this play out. When we "onboard" AI agents, giving them access to institutional and tribal knowledge, the results improve dramatically. Over the next year, we expect to deepen this, providing more organizational context and watching AI excel not just as a general-purpose tool, but as a knowledgeable member of the team.

Building Forward

A few years ago, "company of Builders" was a vision. Today, it's how we work. Our Head of Support builds his own tools. Our Data Analysts build automation pipelines. Our legal team builds review workflows. Our designers ship features. None of them waited for permission to get started. They saw problems, and they built solutions.

The technology will keep evolving, better models, smarter agents, deeper context. But the foundation that makes all of it work isn't technical. It's a culture where everyone is expected to build, empowered to build, and supported to build safely. That's what we've been building for. And we're just getting started.

If this is the kind of environment where you'd thrive, we're hiring.