Caylent Services
Agentic Engineering and Enablement
Turn AI tool use into a governed, compounding capability. Delivered by ACE, Caylent’s Anthropic Consulting & Engineering practice.
Accelerating Secure Enterprise AI Deployments on AWS with Caylent
Generative AI
At a glance
Glean is the context platform for enterprise AI, connecting company knowledge, people, decisions, and workflows across systems to give employees and AI agents the context they need to get work done. As more of the world's largest enterprises choose Glean, many customers want to deploy it in their own AWS accounts—keeping sensitive data within their environment. Caylent embedded a Forward Deployed Engineering pod with Glean's team to build reusable onboarding patterns, creating a secure, scalable deployment path that helps AWS-based enterprises get up and running on Glean faster.
Solution Implemented
→ Embedded Forward Deployed Engineering pod: cloud architects, forward deployed engineers, and a project manager working alongside Glean's teams to support customer deployments.
→ Repeatable deployment harness: started with Glean's most demanding financial-services deployment and turned what the FDE pod learned into a reusable harness of agents and workflows built with Glean. It combines a security design that gives customers' security teams confidence in approving each deployment, the Caylent-built SCP Checker agent that flags blocking account policies before work begins, and private networking and connector fixes backed by customer-facing documentation and training, so each new deployment starts from a proven pattern.
Outcomes
→ Accelerated customer adoption: Streamlined enterprise deployments through repeatable security and infrastructure configurations.
→ 14 enterprise accounts supported with dedicated engineering expertise, and security reviews that start from a proven pattern instead of a blank page.
→ A global semiconductor enterprise live in production on a fully private AWS network, starting with 1,200 sales seats and engineering seats next in line.
→ Sales-productivity agents built with that customer, so account planning that used to take hours now runs in seconds.
→ Recognized as Glean's Architecture Partner of the Year in its 2026 Partner Awards.
Glean started with a simple problem: people could not find the knowledge they needed to do their jobs. Founded in 2019 by Arvind Jain and a team of engineers from Google, Rubrik, and Facebook, Glean began as an enterprise search platform and has grown into the leading context platform for enterprise AI, connecting company knowledge, people, decisions, and workflows across systems to give employees and AI agents the context they need to get work done. Each year, more of the world's largest companies choose Glean, and its Fortune 500 customer base has roughly doubled year over year.
That growth brought a new kind of demand. Many of Glean's largest customers ran on AWS and wanted Glean deployed in their own AWS accounts, keeping sensitive data within their security perimeter while retaining control over security and costs. Every deployment came with its own IAM requirements, network design, and security review. Glean needed a secure, scalable way to onboard AWS enterprises faster, so Caylent embedded a Forward Deployed Engineering pod alongside Glean's teams to help accelerate and standardize these deployments.
Business Challenge
Growing Fast Meant Making Every Deployment Repeatable
As more enterprises adopted Glean, the architecture they asked for most often was also the most demanding: a self-hosted deployment inside their own AWS environment, behind their own guardrails. Each deployment had to pass a rigorous internal security review and integrate with the customer’s existing AWS architecture and controls. After over a year of AWS deployments, Glean identified opportunities to further streamline deployment processes across complex, customer-specific AWS environments.
As an added complexity, the team needed to future-proof their processes for enterprise deployments in regulated industries, which are rarely a simple install. Strict security policies can require additional validation before deployment. Firewall rules built to protect the perimeter can interrupt the sign-in flows teams depend on. Automated setup can break on region and permission mismatches, and standing up a compliant sandbox can take weeks. Private networks, security proxies, and custom domains add review at every step, and each customer's approval process is its own. These requirements underscore the importance of deployment approaches that can accommodate each enterprise's unique security and infrastructure needs.
Solving these one account at a time would not keep pace with Glean's growth, and enterprise customers did not have the internal capacity to clear this kind of complexity. Glean needed a repeatable, secure model to accelerate enterprise deployments on AWS, supported by engineers fluent in both AWS and Glean who could work directly with each customer's environment.
"Building production-grade agentic systems requires deep expertise in both AI and the underlying cloud infrastructure — and Caylent brings both. Their team helped us move from concept to deployment on AWS AI services at a pace that matched our ambition. As Glean helps enterprises put AI to work through secure, context-aware experiences, partners like Caylent who know how to build on AWS at speed are essential to that mission."
Zubin Irani
VP of Partnerships
Solution
Caylent embedded a Forward Deployed Engineering (FDE) pod alongside Glean's teams to support customer deployments. The FDE pod brought together senior cloud architects, forward deployed engineers, and a project manager, working side by side with Glean's own field team.
Start With the Toughest Deployment, Then Repeat What Works
The engagement ran in two stages. In the first month, Caylent’s FDE pod focused on Glean's most demanding deployment, a large financial-services customer with especially strict security and network requirements. Everything learned there became a repeatable approach for future deployments. Following this engagement, the FDE pod expanded to support additional enterprise deployments, so each new customer's security questions could be answered with proven solutions delivered by Caylent instead of starting from scratch.
Security Designed to Contain Risk, Not Just Limit Access
Caylent’s FDE pod standardized a security design that treats each Glean deployment as a separate protected environment, connected to the customer's network through controlled links and governed by layered AWS security policies. The design went beyond just controlling who has access to the system; it also contained the impact if something went wrong. This gave customers' security teams greater confidence in approving each deployment. The FDE pod also built the SCP Checker agent, which spots account-level security policies likely to block a deployment before work begins, helping teams proactively identify and address potential deployment blockers. The tool is now owned jointly by Glean, and includes customer-facing documentation that enables customers to this process.
Solve It Once, Reuse It Everywhere
A typical Glean deployment in this engagement runs inside private networks, with Amazon Bedrock supplying the AI models and AWS security services handling encryption, access control, and web protection. Caylent’s approach has proven effective in fully private environments with no direct internet access, and it produced fixes that Glean now reuses across customers, from connecting data sources privately to resolving the policy and network conflicts that had previously required additional configuration.
Outcome
Faster Deployments, Fewer Stalls, Proof in Production
Since the engagement began, Caylent’s FDE pod has supported 14 enterprise accounts with hands-on engineering support, and Glean has streamlined enterprise deployments through repeatable security and infrastructure configurations.The blockers that once held deals up before work could start now follow a proven pattern, and the FDE pod's fixes carry over from one deployment to the next.
Proof Point: A Global Semiconductor Enterprise
One of the world's largest semiconductor companies shows what this model looks like in production. Caylent's FDE pod implemented Glean inside the customer’s AWS environment, setting it up on a fully private network and working through the company's security controls so Glean could use Amazon Bedrock safely. The team also delivered a custom mobile app, connected Glean to the company's core systems across Salesforce, Jira, and Confluence, and tuned how documents are indexed, drastically cutting the amount of indexed content to help control cost.
Glean launched at the semiconductor company with 1,200 licenses for the sales department, with the engineering department next in line. On top of that foundation, Caylent's engineers worked hands-on with the customer to build a first wave of sales-productivity agents: a Customer 360 and account-planning agent, an RFP response agent, and a meeting-prep agent. The Customer 360 agent pulls from Salesforce, Jira, Slack, Snowflake and Databricks to draft account plans and update records itself. Account planning that used to take hours now runs in seconds across all of the semiconductor company's accounts.
A Partnership Built to Scale
The semiconductor company is one example of a broader partnership. Caylent's FDE pods continue to work with Glean across its enterprise customers, bringing AWS and Amazon Bedrock expertise to the complex deployments that are hardest to get right. Many of Glean's customers already run on AWS, and that overlap is where the two teams add the most value together: Glean brings its context platform for enterprise AI, connecting knowledge, people, decisions, and workflows across systems, and Caylent brings AWS experience applying AI in enterprise workflows. Together, Glean and Caylent help enterprises move AI adoption beyond pilots and into production. As part of the Glean Partner Network, Caylent was recognized as Architecture Partner of the Year in Glean's 2026 Partner Awards, and the two teams continue to scale this approach to more enterprises.
"Building production-grade agentic systems requires deep expertise in both AI and the underlying cloud infrastructure — and Caylent brings both," said Zubin Irani, VP, Partnerships at Glean. "Their team helped us move from concept to deployment on AWS AI services at a pace that matched our ambition. As Glean helps enterprises put AI to work through secure, context-aware experiences, partners like Caylent who know how to build on AWS at speed are essential to that mission."
AWS Services Used
Each Glean deployment in this engagement runs inside the customer's own AWS account, set up with Terraform and AWS CloudFormation and deployed through AWS CodeBuild. Amazon EKS hosts the application in a dedicated VPC, with Amazon ECR supplying signed container images and Amazon S3 and Amazon RDS handling storage. Services such as AWS Lambda and Amazon SageMaker are added depending on which Glean features a customer turns on. Amazon Bedrock supplies the Anthropic Claude models behind Glean's assistant and agents, and Caylent also analyzed Bedrock prompt caching to keep costs in check.
Security and networking carry much of the weight. AWS Transit Gateway connects Glean's environment to the customer's network, AWS PrivateLink and VPC endpoints keep traffic off the public internet, and Amazon Route 53 Resolver handles private DNS. AWS IAM roles, permission boundaries and service control policies define what each deployment can do, AWS KMS and AWS Secrets Manager protect keys and credentials, and AWS WAF guards web traffic. Amazon CloudWatch gives teams visibility into how each deployment is running.
