Connecting Salesforce to Amazon Bedrock Knowledge Bases takes an afternoon. Making sure the assistant does not read people records they were never allowed to see takes longer. A practical walkthrough of the connector, the permission model it quietly drops, and the two architectures that actually hold up.
Most customer sentiment analysis pipelines run perfectly and still produce a number nobody should act on. Here are the failure families that cause it, from labels stamped on the wrong message to averages taken over a scale that was never numeric, plus a pipeline shape and a legal boundary worth knowing before you ship.
Most Salesforce AWS integrations don't fail loudly, they drift. This guide covers the four integration paths between Salesforce and AWS, when to use each, the silent failure mode hiding in every one of them, and the reconciliation job you have to build yourself.
The demo works, then someone sees records they shouldn't. Nothing errors. Here's how to build a Salesforce AI assistant with Amazon Bedrock that respects your sharing model: which doors AWS and Salesforce just closed, why identity propagation is the failure mode that bites, and how to shape the tool surface so the agent can't wander.
Everyone argues this one on data volume, and volume is the argument that matters least: CRM data is small enough that both architectures handle it comfortably. What actually decides data lake vs data warehouse for CRM analytics is how much point-in-time history you need, how fast the schema churns, what shape your queries are, and who is going to maintain the thing. Includes a decision procedure you can run in an afternoon.
Agentforce and AWS wire together in four standard patterns, and every one of them has a point where Salesforce's guarantees stop and yours start. This traces a single request across each boundary it crosses, covers the Trust Layer default most write-ups get wrong (LLM data masking is disabled for agents), and sets out what changes the moment a callout lands in your own account: retention, audit trail, and user identity that does not travel.
A SaaS admin changes a field and your pipeline stays green while the numbers drift. A practical guide to AWS Glue Data Quality for SaaS sources: where to run the checks, why nested payloads need flattening before DQDL can see them, which rule catches which failure, and the dynamic rules that pass silently because they have no history yet.
Most API pipeline failures are green DAGs producing incomplete data. A practical guide to running Apache Airflow on AWS for SaaS and API extraction: choosing between MWAA provisioned, MWAA Serverless and self-managed, the pool setting that silently stops throttling when you go deferrable, retry and pagination design, secrets handling, and the four cost lines that actually move.
Asked how many open opportunities an account had, the assistant said five. It had nineteen. Retrieval returned five chunks and the model counted them. CRM data is structured and RAG is built for text, so the preparation work is mostly deciding which questions are retrieval and which are queries.
"It works" and "it's ready" are different claims. Integrations rarely fail on launch day; they fail in week six, when the token expires and someone adds a field. A readiness checklist for Salesforce AWS integrations, with a pass test on every item.