Tony Capecci is Building Practical Pathways for Legal Innovation

In our Innovation Spotlight series, we interview innovation leaders about how they approach their jobs and how they’ve implemented and deployed Harvey.

In this edition, we chat with Tony Capecci, Director of Practice Innovation at Haynes Boone, an AmLaw 100 firm.

A technologist at heart with more than 20 years in legal, Tony has led teams across the full spectrum of legal applications and now drives AI and innovation strategy at Haynes Boone. His focus is on building practical frameworks, scaling adoption, and creating client-facing value through modern technology. He also volunteers with the International Legal Technology Association (ILTA), where he serves on the Talent Council, a group that helps guide board elections and fosters the next generation of legal tech leaders.

Innovation is about being willing and open to trying new things. It’s easy to fall into patterns of doing things the same way, but in today’s world there are constant opportunities for efficiency and improvement. We owe it to ourselves to explore those opportunities so we can spend more time on the parts of our work we enjoy, and our clients should always feel that they are receiving the greatest value we can provide. For me, innovation means stepping outside of your comfort zone to grow, to stay fulfilled, and to deliver better outcomes.

I love solving conventional problems in unconventional ways. I’m a duct tape and popsicle sticks kind of guy, which has carried into how I approach innovation. During the 2020 lockdown I started learning small electronics repair and restoring vintage video game systems. The work was detail-oriented and required patience, and it reminded me how much creativity comes from rolling up your sleeves and experimenting.

I’ve also pushed myself to take on things that once intimidated me. Flying was one of them. I told my wife that after enough time in PC flight simulators I could probably fly a real-life Cessna 172. For my birthday she surprised me with a lesson, and I did it! I took off, circled the city of Chicago, and landed. That experience cemented a belief for me: Fear usually comes from not understanding, and the best way through it is to just learn the thing.

In my professional life, these lessons matter. Innovation often asks us to work with new tools and untested approaches. It can feel uncomfortable at first, but whether it’s a big leap or a small step, diving in and gaining understanding is how progress gets made.

In my more than 20 years in legal, I had never seen law firms at the cutting edge of technology — until now. It feels like all that patience has finally paid off. It’s an incredibly exciting time to be in legal innovation because the pace of progress is unlike anything I’ve seen before.

Our attorneys at Haynes Boone are generating new ideas every day, and we need reliable ways to capture those ideas and modern, scalable delivery models to bring them to life. In many ways, it feels like a greenfield opportunity. The weeks, months, and years ahead will bring rapid growth as legal tech continues to evolve.

Generative AI makes this even more exciting. I think of it like a ball of clay. You can roll it into a simple sphere, or you can sculpt something much more intricate. The outcome depends on the time and effort you put in, and that is exactly the role of innovation teams today: shaping that clay into something meaningful for the business and for our clients.

I first used Harvey when it was a single text box on the screen that said, “What do you want to do?” In just a couple of years it has grown into a very compelling application. Harvey listens to its customers and rapidly develops the features they ask for. It’s rare to see a company deliver on its roadmap with such consistency and speed. The shipping velocity is incredible, and the level of engagement has been outstanding. Whether it’s brainstorming training initiatives or sharing updates on their roadmap, Harvey has been a true partner to us as we continue to round first base in the game of AI.

One of the most powerful features is the Workflow Builder. I think of Harvey in two ways. The Assistant is where you can upload documents, enter a query, and get answers — simple and effective. The Workflow Builder is where the real money is. Attorneys are extraordinarily busy serving clients, and they don’t always have time to become detailed prompt engineers. The Workflow Builder gives our innovation team the ability to create that “easy button” for lawyers. It hides the complex prompt engineering behind the scenes so the attorney only needs to validate the output, not build the workflow. That combination of simplicity and depth is what makes Harvey such a strong fit for our firm.

We are seeing strong adoption across the firm, from Partners to Associates, with champions in every practice group driving engagement and helping others explore Harvey’s capabilities. Common use cases include proofreading for spelling and grammar, summarization, and list creation. Attorneys are also using Harvey as a “sparring partner” by testing arguments, generating counterpoints, and suggesting revisions. Beyond that, we see persona-based engagement where complex concepts are distilled into narratives for clients, as well as data extraction and analysis that make large reviews more manageable.

Another area of growing interest is Harvey for Word, which has become more interesting since its recent updates. Attorneys spend so much of their day in Word that being able to access Harvey without leaving that environment is a real advantage. The playbook capability is particularly interesting, as it allows attorneys to organize and reuse their best workflows.

I still enjoy one-on-one training, because that’s usually when I see attorneys have their eureka moments. Those moments matter for change management, because they are the point where trust starts to build. We focus on “show me the money” use cases — attorneys need to see something that is useful and compelling in their day-to-day work before they will adopt.

I often say AI stands for It’s About Iteration. When an attorney is frustrated with an output, I like to sit down with them and do a little prompt engineering. Nine times out of ten, we can improve the result right in front of them. That shift in perspective is the eureka moment, and it helps turn initial skepticism into momentum.

Beyond one-on-one training, we also run group sessions focused on both transactional and litigation practices, and we are adding bespoke workshops for specific practice areas. To scale this approach, we are expanding our innovation delivery team. By combining internal business analysis, product ownership, and engineering skills, we can keep delivering those eureka moments across the firm on a regular basis. This structure ensures that adoption doesn’t just rely on isolated success stories, but becomes part of a consistent process.

One example is a recurring project where attorneys need to extract and analyze data across a large set of documents in order to identify patterns and draw insights. In the past, this required months of manual review. With Harvey, we built a workflow that handles the first pass of data extraction and organizes the results into a structured format that attorneys can review and refine.

What once took months can now be completed in a matter of hours, freeing up attorneys to focus on the higher-value parts of the work. It took about a month of prompt engineering (and testing results with different models) to dial in the workflow to the level of accuracy we wanted, but the impact has been substantial.

Another strong use case comes from our finance practice. We designed a workflow in Harvey that extracts specific data points from documents and compiles them into a checklist. Harvey produces the first draft automatically, which then goes to a partner for review and approval. During the pilot, we went through several rounds of prompt refinement to improve accuracy, and today the workflow is highly reliable. It has reduced the time needed to prepare these checklists and created a more consistent end product, while always keeping a human in the loop.

These examples illustrate how meaningful results come from iteration. Developing these workflows was not an overnight success — they took careful testing, refinement, and adjustment. But once established, the payoff is clear. I believe the juice is worth the squeeze when it is something you need to do more than once.

There’s a lot of conversation around how AI will transform legal services, and I believe much of it. But for me, real success will be when we look back in a few years and see how deeply these tools have been woven into everyday practice. I imagine a moment where we take stock of all the AI-driven workflows, automations, and efficiencies in place across the firm and realize just how much progress has been made, often quietly and step by step.

Success also depends on having the right visibility. It is very beneficial when these tools provide granular usage data that we can enrich with our internal systems and dashboards. That allows us to understand exactly how practice groups are engaging, identify champions, and spot the people who may need a little more support. With that level of insight, we can fine-tune training and shape deployment strategies to maximize adoption.

I think of success as taking small, steady steps rather than trying to do everything at once. The old “crawl, walk, run” adage applies here. It is okay to crawl and walk for a while until you are ready to run. Keep building reliable workflows, measuring adoption, and using data to guide efforts in order to create a foundation that scales. That’s what long-term success with GenAI will look like.

Innovation comes down to two things: ideas and delivery. You can’t have one without the other. If you want to deploy GenAI effectively in a law firm, you need a framework for execution. Without it, you risk staying in experimentation mode indefinitely. At some point, the rubber needs to meet the road.

It’s tempting to want to do everything at once, but the better path is to take small, deliberate steps. Listen closely to your attorneys, and do not overpromise on what these tools can do. Many people have had disappointing first experiences with generative AI. The way forward is to show meaningful results that matter to them in their moment of need.

My main takeaway for both attorneys and innovation teams is simple: Do not give up. Progress happens through persistence, small wins, and delivering value consistently. That is how you build long-term trust in these tools.

The impact will be simple but powerful: remarkable efficiencies in the day-to-day work of attorneys, which translate into higher quality output for clients. The real opportunity comes when firms combine GenAI with their collective knowledge. That knowledge, organized and enriched by AI, becomes more accessible, more actionable, and more valuable in the hands of attorneys.

In the end, GenAI will change how legal work gets done, and it will change the scale and speed at which attorneys can deliver meaningful insights and advice. That shift has the potential to redefine the client experience and raise the bar for the entire profession.