Overview of Solving integration woes with a hackathon
This episode of the Stack Overflow Podcast explores how Adobe approached a post-acquisition integration challenge by running a focused internal hackathon with Semrush. Guest Meryl Blanchet, Director of Engineering for Adobe Brand Visibility, explains how the two teams used the event to align culturally, combine complementary AI visibility products, and ship customer-facing value quickly—rather than spending months on a traditional, heavyweight integration plan.
Why Adobe used a hackathon for integration
A customer-value-first approach
Instead of starting with deep infrastructure rewrites, the teams began with a simple question: what value can we unlock for customers right now? The hackathon was designed to:
- Bring Adobe and Semrush engineers together quickly
- Solve concrete product problems
- Deliver something usable within a very short time frame
- Focus on the first GA release and customer zero, adobe.com
The goal
The north star was to combine strengths from both sides to build the best AI/geo brand visibility product in the market, while keeping the work practical and shippable.
How the two products complemented each other
Semrush’s strength
Semrush brought a highly data-centric product, focused on:
- LLM visibility data
- Prompt and topic tracking
- Enterprise SEO/geo analysis
- API-first access to data
Adobe’s strength
Adobe’s LLM Optimizer, built from the perspective of enterprise CMS and content workflows, contributed:
- On-site telemetry
- CDN logs
- Adobe Analytics / Customer Journey Analytics signals
- Actionability for marketers and content teams
The combined vision
The integration aimed to blend:
- Off-site signals from Semrush
- On-site signals from Adobe
- Marketing actionability for content creation, optimization, and response
Biggest integration challenges
1. Culture and operating model
The teams had different repositories, workflows, and engineering norms, but the transition was smoother than expected. One notable success: a Semrush engineer submitted the first pull request into an Adobe repository on day one.
2. Tech stack and UI integration
Adobe reused the existing LLM Optimizer UI shell and infrastructure so the new customer-facing features could fit into Adobe’s generally available environment. This also made it easier for Semrush engineers to onboard and contribute through Adobe’s CI/CD setup.
3. Data model differences
A major conceptual difference was how each company framed the product data:
- Semrush: centered around domains
- Adobe: evolved toward a more brand-centric model spanning multiple domains and signals
The teams preserved both approaches, but unified them under Adobe Brand Visibility’s brand-centric model.
What the hackathon produced
Delivered quickly
The hackathon wasn’t open-ended brainstorming—it was tightly scoped to a few high-priority workstreams. Outcomes included:
- Integration of Semrush AI visibility capabilities into Adobe Brand Visibility
- Early customer-facing functionality on adobe.com
- Infrastructure and identity alignment for Semrush within Adobe
- Product planning support from PMs and PMMs
A standout feature: prompt strategy agent
One of the most important outputs was a prompt strategy agent. It helps customers:
- Identify relevant prompts and topics to track
- Combine market signals with on-site signals
- Get recommendations for what to monitor and why
- Reduce the manual effort needed to set up and maintain visibility tracking
Lessons learned
Engineer creativity works best with constraints
Meryl emphasized that hackathons are most effective when they are:
- Goal-oriented
- Customer-driven
- Focused on a clearly bounded set of outcomes
The team intentionally avoided overloading the week with long-term wishlist items.
Domain expertise matters more than ever
A key takeaway was the value of hiring domain experts—especially in SEO/geo and related areas. These experts help engineering teams:
- Better understand customer problems
- Validate feature relevance
- Translate raw data into practical product decisions
Fast-moving markets require flexibility
Because the AI search and brand visibility landscape is changing rapidly, the team keeps a close eye on:
- New LLM behaviors
- Market-specific trends
- Feature relevance over time
They also stay willing to remove or deprecate features if they stop delivering value.
Broader takeaways
- Hackathons can be a serious integration tool, not just a novelty
- Customer zero matters: shipping to adobe.com helped validate the combined product quickly
- API-first products are easier to integrate
- Identity, branding, and data models are often the hardest parts of post-acquisition product work
- Human review still matters, even in AI-assisted workflows
What’s next
The team is already planning a follow-up hackathon in Semrush’s office focused on new two-pick capabilities. The broader roadmap includes:
- More feature validation with early-access customers
- Deeper data combination across Adobe and Semrush signals
- Continued work on prompt tracking and prompt strategy
- Further productization after the initial GA release
Notable quote-level insights
On hackathons
- Hackathons are useful when they’re not just about exploration, but about delivering concrete outcomes fast
- The best work happens when engineers know the goal and are given room to solve it creatively
On product building
- “If you don’t have empathy for the end user, you can’t make a good product.”
- Validating early with real customers is essential in a market moving this quickly
Mentioned resources
- Adobe Brand Visibility preview:
https://play.bv.now - Stack Overflow podcast shoutout: Patty Jane, for the question about setting a macOS PATH environment variable permanently