At AstraZeneca, the interesting number was two. The pharmaceutical company wanted cloud applications and older systems to cooperate without assembling a large integration department. After an initial rollout, a two-person central services team helped teach people across countries and business units to build their own connections. A little team was becoming a school.
- The job: connect data, applications, APIs and AI agents.
- The method: visual pipelines assembled from reusable connectors called Snaps.
- The useful lesson: a small team can spread capability through training and shared rules.
- The buying question: can your team own the workflows after the demo ends?
Two people, hundreds of connections
In SnapLogic’s September 2016 account of AstraZeneca’s rollout, the first sales and marketing integrations reached production within four months. The company reported 100 integrations in production within six months, more than 300 subsequently in use, and over 100 citizen integrators around the world. Those figures describe that early program, rather than AstraZeneca’s current estate.
The surprising part is the organizational design. The central group trained distributed builders instead of personally constructing every connection. Its contribution was expertise that could travel. A business user with a specific problem no longer had to wait for every detail to pass through the same narrow doorway.
“spend an hour with us, and we’ll show you the value of SnapLogic.”
Don Shelly, Solution Architect at AstraZeneca · 2016
That is an appealing bargain: an hour of attention in exchange for a new capability. It also suggests a better question than whether a tool is “easy.” Easy for whom, doing what, with which rules? A screen can be friendly while the process behind it remains a committee with no chair.
The cloud needed an introduction
SnapLogic began in 2006. Co-founder Gaurav Dhillon had already helped build Informatica, an earlier integration business. In a 2011 interview, he described the new problem: enterprises were acquiring cloud applications that had to connect with existing systems, and the integration technology itself needed to accommodate that change. He became SnapLogic’s CEO in 2009.
The early company spoke a different technological dialect. A 2008 Salesforce solution pack was free and licensed under GPL v2. Today’s business sells enterprise subscriptions and AI orchestration. The recurring concern is recognizable across those eras: buying an application does not make its information available wherever the business needs it.
Siemens Digital Industries supplies a more physical example. Its growing data volumes were straining existing infrastructure. According to a December 2020 customer announcement, Siemens combined SnapLogic with Kafka streaming to provide product master data to more than 80 regional sales teams. That data supported activities including real-time quotation generation. The team reported encouraging results after a successful pilot.
Here the thing that faltered was the old data arrangement under increasing volume. The response was a changed architecture, tested before expansion. A quotation is a small commercial promise; getting the product information wrong can make it an expensive one. Integration earns its keep in these ordinary transactions.
A box of Snaps, with grown-up responsibilities
SnapLogic’s library of more than 1,000 Snaps supplies prebuilt connections to applications, databases and other services. A builder arranges components on a visual canvas to form a pipeline. Instead of writing every interaction from scratch, the builder configures reusable pieces and the transformations between them. Developers can also extend the connector library.
Its platform spans several kinds of integration: moving and transforming data for analysis, connecting applications for business processes, and managing APIs that expose functions to other software. Cloud and on-premises deployment matter because a new SaaS application may still depend on a database behind the customer’s firewall. Most businesses inherit their architecture one purchase at a time.
The workflow still needs credentials, clear rules, testing and an owner.
A useful onboarding pipeline might read an approved employee record, map fields and pass the right information to downstream systems. A data team might move operational records into a warehouse. These are examples of the pattern, not a promise that every endpoint or approval rule is already packaged.
The published operating workflow still includes validation, testing, deployment and monitoring. Low-code changes how people construct the work. Someone must still decide which record wins when two systems disagree, what happens when a request fails, and who receives the alert. A pretty diagram is poor company at three in the morning.
What the subscription buys
SnapLogic sells Essential, Professional and Enterprise One packages. Its current pricing page advertises unlimited pipelines and data movement within its package approach; premium connectors and features are additional buying decisions. Exact prices require a quote. Services include architecture advice, migration and implementation help, while OEM arrangements let other software providers embed the technology.
The sensible cost comparison includes the subscription, chosen add-ons, implementation effort and continuing operations. For AI workloads, customers separately procure supported model and vector database services, as the AgentCreator product FAQ explains. Unlimited movement on the integration platform does not settle every bill generated by the systems it connects.
This is a crowded market. MuleSoft combines integration with API development and management. Workato also offers business automation, low-code agents and governed tools. SnapLogic’s pitch is the combined scope of its platform and its visual pipeline approach. Whether that combination fits depends on the buyer’s applications, skills and existing investments; connector counts alone cannot settle it.
The company’s scale is substantial enough to make the stakes plain. In August 2025, SnapLogic reported passing $100 million in annual recurring revenue as Brad Stewart became CEO and Dhillon moved to Chairman. ARR measures recurring contract value, rather than audited annual revenue. Its $165 million December 2021 financing valued the company at $1 billion at that time.
Now the connector has an AI colleague
SnapGPT and AgentCreator have different jobs. SnapGPT assists the people constructing integrations. AgentCreator builds LLM-powered agents and applications that use connected systems. In July 2026, SnapLogic expanded SnapGPT across planning, generation and operations. The distinction matters: helping a builder design a workflow differs from letting an agent act through it.
The documented HR Q&A example gives the idea a modest, practical starting point. An employee asks about policy; the workflow uses information from an employee handbook. The supporting machinery prepares and retrieves relevant material for the model. There is a great deal of data work hiding inside a seemingly effortless answer.
An AWS technical account of AgentCreator describes using Amazon Bedrock with enterprise knowledge and recommends beginning with a bounded departmental project. SnapLogic’s technology partnerships also span Snowflake, Microsoft and Google Cloud. Those relationships help the platform meet customers where their data and applications already live.
Action adds a sharper question: whose authority does the agent possess? SnapLogic’s April 2026 AI Gateway and Trusted Agent Identity announcement describes centralized access controls and carrying the initiating user’s identity through the integration to the backend system. The practical aim is to keep an agent’s actions tied to the permissions of the person requesting them.
Meanwhile, old integrations still need attention. On September 22, 2026, SnapLogic expanded Intelligent Modernizer with an agentic engine for analyzing legacy workloads and generating pipelines, alongside broader platform coverage and governance controls. The AI future includes quite a lot of archaeology: discovering what yesterday’s software actually does before replacing it.
Borrow the teaching model
SnapLogic’s careers material emphasizes learning, creative problem solving and teamwork. The customer story at the beginning offers a concrete expression of those ideas: expertise becomes more valuable when people can pass it on. The platform lowers some construction costs; the training model increases the number of capable builders.
My reading of these cases is that the copyable move comes before a sweeping rollout. Pick one useful process. Name its owner. Check its data, permissions and failure handling. Measure whether it saves work. Turn the successful design into a reusable pattern, then teach someone else to operate it. A small central team should remain responsible for standards and support.
That approach depends on willing learners and agreed responsibilities. A few simple connections may not justify a broad enterprise platform. Unresolved data ownership, unsupported system behavior or excessive AI permissions can swallow the gains. The rewarding part of SnapLogic’s proposition is that it can give more people a way to build. The company buying it must give them something sensible to build together.