Discover more from The API Economy
Merge Blueprint: Automating Integration Builds With AI
Merge is setting the precedent for how companies should leverage AI to drive real impact for their customers & users.
Welcome to the latest edition of ‘The API Economy’ — thanks for being here.
To support The API Economy, be like one of the 3,644 early, forward-thinking people, who subscribe for monthly-ish insights on the API economy, Web3, AI, and more emerging trends.
Setting an AI precedent for technology companies
Hello everyone 👋,
Today’s edition of The API Economy is brought to you in partnership with Merge.
While it’s commonplace by now to hear about companies completely pivoting to AI or launching full-blown AI projects, the Merge team took a more pragmatic approach.
With Blueprint, they seem to have found the perfect recipe of:
Identifying a problem AI can help with.
Implementing an AI solution in a way that drives real impact.
Making the solution in a form that makes the most sense.
For Merge, the result is an internal product that speeds up their team’s velocity to build and takes much of the manual work out of their development process when building out an integration.
The tool leverages AI to analyze API documentation and automatically infer connections between APIs and Merge's Unified APIs. An engineer then ensures the work meets their quality standards.
Recently, I had a conversation with Merge's co-founder, Gil Feig, to learn more about how Merge is revolutionizing integration development through AI.
In this article, we’ll be taking a closer look at why you don’t need to be an AI company to benefit from building with AI, what API-first companies can learn from Merge’s AI approach, and more about how Blueprint works.
Integrations connect software
First, let’s start with some context on why Merge exists.
Integrations connect tools, applications, or data through application programming interfaces (APIs). These integrations can be costly to build and difficult to maintain, but they’re crucial for businesses that want to drive growth by creating more frictionless experiences for their users.
SaaStr reports that top SaaS companies have an average of 350 integrations and the biggest names in tech have well over 1,000 integrations.
Clearly, integrations matter in SaaS. But why?
Without integrations, products and data are siloed from the rest of the SaaS ecosystem. It makes stand-alone products a one-trick pony that isn’t able to benefit from the tangential best-of-breed products.
Users need products to work together to reduce manual work, build custom workflows, and let software do what it is meant to do — make life easier.
“If the number of potential connections between APIs increases exponentially as more are added, companies have a near-infinite ability to create unique chain-link systems of coherent actions out of the existing primitives. Instead of a linear value chain in which using commoditized components in each step limits the number of places left to differentiate and create value.”
Integrations provide access to the data and platforms that power impactful use cases for SaaS apps. With them, companies have the ability to do things like:
Pull in accounting data to make headcount planning easier for customers.
Forecast a customer’s future revenue and upsell potential based on deal flows from a CRM.
Create searchable databases of a company’s files and folders.
These integrations have a massive impact in SaaS but unfortunately, many start-ups still struggle to build and maintain their own integrations.
Let’s take a closer look at why this process can be so challenging for companies to prioritize.
Building integrations is harder than it looks
Developing integrations from scratch can be a daunting and time-consuming task for developers. The process typically involves meticulously studying API documentation, deciphering complex data structures, and manually mapping connections between different systems.
This manual work is often prone to errors, requires extensive coding, and demands significant effort to ensure compatibility and seamless functionality. It isn’t low stakes, either: if an integration is unreliable or breaks down frequently, the consequences for a business could be massive — not to mention expensive.
That said, product managers often face a challenging balancing act when deciding to allocate resources between building integrations and enhancing their product's core competencies. Integrations can provide significant value to customers and to the internal team, but focusing energy here often means neglecting the product itself — which can also put you behind your competition.
Building robust integrations requires significant time, technical expertise, and often collaboration with other software providers — resources that might otherwise be invested in enhancing the product's primary features.
These complexities, combined with the pressures of managing a product roadmap within tight timelines and budgets, mean that some companies feel stuck with broken, unreliable integrations (if they’re lucky enough to have them in the first place).
Merge is here to help product teams to focus on their core competencies while leaving their API strategy to the people who know it best.
Packy M frames the opportunity when discussing how companies are increasingly able to build nearly everything non-core through APIs:
“What people call “strategy” is often a bunch of fancy words and falsely precise goals that obfuscate the core purpose of the business: serving customers in a differentiated way. Judiciously using APIs where possible gives companies strategic clarity and the ability to solve their customers’ problems in a way that only they can.”
There are two ways to use APIs to solve your customer problems.
Build each individual integration in-house.
Integrate with Merge once and get access to hundreds of integrations.
If you take on integrations in-house, some companies struggle to ship a handful of integrations a year. But since we last spoke to Merge 18 months ago, they have added 4 new integration categories (including File Storage which they shipped this week), added hundreds of integrations, and the team shows no signs of slowing down.
Merge is hands down the best in the business when it comes to integration strategy & execution and it’s hard to make the case that any internal team can ship the way they do.
But why would development teams even want to try to take on integrations themselves? Development teams have enough to focus on already just to make their core products and features the best they can be without worrying about integration strategy.
“We were worried about what we internally called the Microsoft Word toolbar problem. Now this is hard to explain, because now Word has a ribbon and hides a lot of this stuff. But back in those days, and we talked about it, Word had toolbars, and if you were on a normal 17-inch screen, and you would enable every single toolbar that you could, you would actually only have a centimeter of space for actually typing text, right?
Because Word had a lot of features, and packaged goods software has to make another sale every single cycle that makes perfect sense. But we thought that was a problem for us building Shopify, because e-commerce is – we are the home of these businesses. We are the software that people open up in the morning and that’s their work day, where they spend it in.”
Integrations are almost always something companies should outsource while focusing on their core product components.
Simplifying integrations with AI
Netflix made it possible for you to log in once and stream thousands of films. Merge is doing the same thing with integrations. Integrate once with their API and get access to all of their integrations.
To briefly recap what Merge does from our previous essay: Merge’s proprietary Unified APIs allow companies to focus on building their own product, and not on the costly, time-consuming task of developing third-party integrations with their customers’ software of choice. By integrating once with Merge’s Unified API, customers can instantly access hundreds of integrations.
Merge has transformed the way that B2B companies develop products. Instead of including complex integrations in their product roadmaps, builders can focus on their product’s core competencies and end-user experience — and leave the mission-critical (yet time-consuming) work of integrations to Merge.
Since its start, Merge has been backed by impressive investors like Accel, NEA, and Addition, raising $75 million so far.
Now, Merge has announced an AI-powered tool called Blueprint to continue growing its long list of compatible integrations. Essentially, Blueprint allows anyone to request a new integration with Merge’s Unified APIs and then automates part of the integration build:
VentureBeat covered the launch and wrote:
“Blueprint allows developers — or even a company’s non-technical employees — to copy and paste a link to the documentation of an API they wish to integrate with into an online form. Blueprint then rapidly reads it, parses it, and translates it into a workable API definition that Merge can use to construct a fast, secure, and easy integration with the developer’s own software.”
As a result, Merge went from moving fast to operating in overdrive.
Here’s how it works:
The user inputs an API documentation URL into Blueprint.
Blueprint automatically generates the API definition (‘blueprint’) to Merge’s Common Models.
Blueprint then starts the build, taking out some of the most monotonous, manual work, like setting up authentication, pagination, endpoints, and matching it up to Merge’s common model format.
The user is given the opportunity to request Blueprint’s output be added to Merge’s Unified APIs. If the integration is already a part of Merge’s integration offering, then the user will be prompted to sign up to get API access and start building into Merge.
Merge will review and finish up the integration — and since the work’s already been started by Blueprint, Merge’s team can release new, customer-requested integrations much faster with customer input.
Why Merge made Blueprint
When it comes to customer-facing integrations, the initial build is just the tip of the iceberg. Once you build an integration, you embark on the never-ending journey of managing and maintaining it (a compounding problem).
An integration going down can bring a product down or render it completely ineffective.
Reliable integrations require a constant dance of monitoring performance, addressing errors, and staying in sync with ever-evolving APIs.
That’s why customers are eager to hand not just the initial build to Merge — but the entire integration lifecycle. As Gil puts it, “At Merge, we know we need to move fast and build fast because companies are placing the fate of their integration strategy in our hands.”
Gil also adds,
“We’re never going to see every B2B company building integrations with every other B2B company, so we need a single language to communicate between them. Then, if all these companies build to one language, they’re essentially building to each other. We want that language to be Merge’s Common Models. This led us to create Blueprint because we were asking the question, ‘how can we make it even easier to just automatically add new platforms and integrations to Merge, and thus, our customers?’”
At its core, that’s exactly what Blueprint is: a tool that will help Merge meet customer needs and build new integrations faster than ever.
Merge is doing what many tech companies aren’t: they are actually using AI to better serve customers. Blueprint allows anyone to request a new integration in seconds (and start the build automatically), meaning Merge can get real-time feedback on what integrations are needed most.
The result? Merge is cutting down integration development time significantly. By automating all of the tedious parts of the development process, the Merge team is able to focus on the important parts of the integration development process like reliability, security, and operability.
And people are taking notice of companies like Merge which leverage AI the right way. Even Elon Musk seems to be a fan of rapid AI adoption like Blueprint:
What companies can learn from Merge
Everywhere you turn these days, tech companies are jumping on the artificial intelligence bandwagon.
So when Merge’s team began investigating opportunities to use AI, they knew they had to have a real use case for it. Gil put it this way:
“We know thin wrappers around ChatGPT aren’t the next big thing. Our goal was to create a product with true value, where AI was simply another tool in our belt to accomplish it. ”
Instead of forcing AI into its product roadmap, Merge built a tool that will massively benefit customers and speed up progress towards Merge’s ultimate goal: to be the integration layer for all B2B companies.
Merge’s founding engineer Henry Baer weighs on the benefits of building Blueprint and how it sets the team up for long-term success leveraging AI:
"Building Blueprint has accelerated our understanding of how LLMs can interpret code, such as API reference docs. We need our API-based integrations to be of the highest quality, so we started with a product that would be useful for our community while also helping us learn its long-term potential."
Clearly, Merge is doing something right. Let’s take a look at 3 lessons API-first companies can learn from Merge’s approach to implementing artificial intelligence.
Lesson #1: Build with AI when it actually makes sense.
Artificial intelligence is incredibly powerful, but it’s also trendy. In the past year alone, dozens of tech companies have fallen prey to the “AI hype cycle” and spent precious resources building some pretty underwhelming tools.
These AI releases might give companies some press or momentary social media engagement, but they’ve rarely been aligned with the company’s actual mission. Some start-ups have even been founded to replicate what ChatGPT…like this AI company in Europe that raised a $100+ million seed round — without a product.
But this doesn’t mean you should ignore AI completely: in fact, to do so would be just as risky. Instead, allow the AI trend to spark creative ideas for better ways to serve your customers. Gil says,
“Every company should think about how AI can enhance what you’re already doing or help launch powerful new products, but don’t force a use case if there’s not one.”
Kieran: “There’s the 90% I think most people are trying to replicate the things we already do (with AI), but that 10% is super interesting which is thinking about what are the things we can do to scale today which could not do previously without this technology (AI).
Bryan: “I get very excited about existing companies as well as net new companies because they now have this new thing (AI). For existing companies, there is this way to deepen your moat and use AI to abstract complexities away to make your product so much more useful than before.”
In Merge’s case, the team decided to move forward with Blueprint because they saw how it could have a lasting impact on the way they serve customers.
Lesson #2: Leverage AI to enhance your product, not as a product itself.
According to Gil,
“When something that’s as game-changing as AI comes about, you have to consider how to use it — but if AI is your product, you probably aren’t doing it right.”
Instead, Gil encourages companies to consider how AI can become a core competency for your business — something that enhances what you already have. How can artificial intelligence actually make your customers’ lives easier and provide tangible value?
“If you spend two weeks building something, anyone else can build it in two weeks. It’s okay to take time here – just put one engineer on it and see what you can do. But don’t just build another thin wrapper on GPT if that won’t serve your customers and provide true value.”
“AI generation is a black box: you can have an expectation of what sort of output you’ll get, but you’re essentially throwing spaghetti at the wall until the algorithm produces something that you like. Unlike the tools we are used to, which work via straightforward inputs and produce straightforward outputs, we’re getting value from AI tools mostly by granularly fine-tuning our incantations to appease the AI – this is a capricious genie, not a well-oiled mechanism.”
Though Blueprint is most powerful as an internal tool for Merge’s team, it came about as a result of the team’s customer-centric approach. Not only does Blueprint all Merge to build integrations faster so they can serve more users and help more businesses — it allows people to request what integrations they want.
As Merge dreams about the future of their AI product development, they’ll continue to focus on enhancing the product, rather than getting distracted by the generative AI hype. For example, Gil hinted at potential future applications for Blueprint, like streamlining the customer onboarding process and supporting the self-healing of API errors.
If you’re considering using AI on your team or in your start-up, try to get creative. Maybe AI could transform your processes and be integrated as an internal tool. Or, maybe there is a true use case for a chatbot of some sort. At the end of the day, don’t be afraid to experiment.
“AI stokes creativity by allowing a company to test ideas rapidly and to do more at scale. Furthermore, it learns from the past, across millions of data points, unlocking innovation quicker than a human could. But AI does not invent; it just predicts, on the basis of past patterns. Marketers invent, and the AI learns what works, for whom, when, and how. Invention requires a culture that values experimentation and risk-taking.
All else being equal, a company’s experimental data is the source of its competitive advantage. Leaders must recognize that a test-and-learn mentality is essential for translating that data from raw material into currency. That means accepting that experimentation has opportunity costs and that by definition, some experiments won’t work. But even failed experiments offer worthwhile lessons.”
Lesson #3: Use AI to supercharge your employees’ impact.
AI can’t fully replace humans, but it has the potential to massively reduce manual labor and allow those team members to focus on more challenging, strategic, and rewarding problems. It has the potential to supercharge the way we all work.
While AI won’t do all the work in software development, it’s clearly trending in a direction that shows that in a relatively short period of time, it will be doing the majority of the work.
AI's ability to automate routine tasks, process vast amounts of data, and generate valuable insights can help startups punch above their weight, accomplishing more with less.
For a small, resource-strapped team, this means that individuals can focus their skills and energy on strategic planning, innovative thinking, and unique problem-solving — the human-centric aspects that truly drive a startup's growth and success.
So ultimately, if your company is considering what to do with AI, a great place to start is by looking for ways to help your team move faster and more efficiently. This is exactly what Merge is doing, and Blueprint is already supercharging its team’s impact and results.
What’s next for Merge?
In this time of technology-led transformations, Merge finds itself at an inflection point. AI could become the crucible where Merge's strategic focus to become the integration layer for B2B SaaS, could be forged into a product of unsurpassed capability and elegance.
By harnessing the robustness of AI, and coupling it with the pervasive need for seamless integration in SaaS platforms, Merge is poised to straddle a critical nexus point. If Merge can surf this wave of AI-integrated solutions, its future could well be a chapter in the annals of technology that future generations will read with awe and inspiration.
As for what that could look like, Gil shared that Merge’s vision for Blueprint includes self-healing and monitoring capabilities that will majorly streamline the resolution of API issues:
“In engineering, everything is proactive — you have to think through endless factors and edge cases to make sure you’re not missing anything. You want to be super confident when you launch. The unfortunate part of APIs is you don’t control third parties, so being reactive is inevitable. AI presents us with the opportunity to be reactive but with the same external reliability as being proactive.”
In the short term, AI will enable Merge to be reactive to issues, even enabling many of those issues to fix themselves quickly. Eventually, this can reduce the time that engineers have to spend thinking through all those edge cases — they’d be able to trust the AI to catch the error and self-heal.
In the long term, when APIs meet AI, they're not just facilitators of software communication anymore, but turn into dynamic conductors, orchestrating complex symphonies of cross-platform operations, enabled by real-time learning and decision-making. Integration makes technology more powerful than ever, not merely by amplifying the capabilities of individual software components, but by unlocking the potential of software as a whole — interconnected, intelligent, and in sync. It builds a network effect for software — the more they are linked, the more valuable they become.
The redefinition of software connectivity patterns Merge is leading has the potential to unprecedented levels of automation, personalization, and efficiency in every field that relies on technology. In essence, AI and APIs could be the magic wand that conjures the next seismic shift in the technology landscape.
In the coming years, I have no doubt that Merge will continue to push the boundaries of AI & APIs. I can’t wait to see what they do next.
Until our next adventure.
Special thanks to Gil & Shensi at Merge, my friend Haley Davidson for copy help & edits, and Mama Schroeder for additional edits (any typos are on them 😊).
Disclaimer: The views in this essay are my own personal opinions and don’t necessarily represent the views of my employer, Merge, those mentioned in this article, or anyone other than myself.