Playbook

Splunk holds the machine data security and operations teams live in, queried through a language most of their colleagues never learn. An agent that turns questions into good SPL widens access and, done carelessly, exhausts search capacity. This guide covers grounded SPL generation, search head protection, role-based access enforcement, alert triage for security operations, and evaluating whether the generated searches are actually right.

AI agents fail in the middle, wait for humans, and run for hours, which is exactly what Temporal's durable execution was built for. This guide covers structuring an agent as a Temporal workflow with model calls as activities, handling human approval through signals, retry and timeout policy for probabilistic steps, versioning prompts and logic safely, and observing what a long-running agent did.

Infrastructure as code is the highest-blast-radius code in the organisation, and an agent that can run apply is a credential that can delete production. This guide covers what a Terraform agent should do, including plan interpretation, drift explanation, policy-checked module generation, and cost estimation, how to keep it reviewable through pull requests, and why apply authority stays with people and pipelines.

Most enterprise AI work is batch: nightly classification, weekly embedding refresh, scheduled evaluation runs, and document backfills. Airflow already runs the data platform's batches, and it can run these too if the model calls are placed correctly. This guide covers structuring AI DAGs, pushing inference to executors rather than the scheduler, idempotent tasks, evaluation as a pipeline, and cost-aware scheduling.

BigQuery is fast, serverless, and billed by bytes scanned, which means an agent that writes careless SQL is both a data exposure risk and a cost risk. This guide covers IAM and row-level security for agent identity, grounding SQL generation in dataset metadata, controlling cost through dry runs and byte limits, and returning results users can verify.

Putting a model inside a Kafka pipeline turns every event into an inference call, which is powerful at scale and ruinous when designed carelessly. This guide covers consumer group design for inference workloads, backpressure when the model is slower than the stream, ordering and exactly-once semantics, schema governance for AI-enriched events, and keeping inference cost proportional to value.

MongoDB's flexible schema is the reason applications choose it and the reason an agent querying it needs care: documents in one collection may not share a shape. This guide covers schema discovery, generating aggregation pipelines safely, using Atlas Vector Search for retrieval alongside operational data, scoping database roles, and keeping writes bounded.

Graph databases answer the questions relational and vector stores struggle with: who is connected to whom, through what, and how far. An agent on Neo4j can exploit that if it generates Cypher safely and understands the graph's schema. This guide covers schema grounding, constrained Cypher generation, combining graph traversal with vector retrieval, and scoping what the agent may traverse.

Redshift sits at the centre of many AWS data estates, with workload management queues and a cluster or serverless cost model that an agent must respect. This guide covers Data API access with IAM identity, row-level and column-level security, grounding SQL in system catalogue metadata, protecting queues from agent load, and returning results users can verify.

Elasticsearch already does the two things retrieval agents need, keyword and vector search, and holds the logs that operations agents reason over. This guide covers hybrid retrieval combining BM25 and dense vectors, document-level security so results respect permissions, safe query DSL generation, and building agents over log and metrics data without overwhelming the cluster.

Databricks holds the lakehouse data and, through Unity Catalog, the governance that makes an agent's data access defensible. This guide covers grounding agents in Unity Catalog metadata, executing queries through SQL warehouses as the requesting user, using Databricks model serving and vector search, and evaluating agents with the platform's own tooling.

Klaviyo holds the customer event stream and the lifecycle messaging for a large share of e-commerce brands, which makes it a strong base for an agent that improves campaigns and a dangerous one for an agent that sends them. This guide covers API and event access, segment and flow analysis, content drafting under brand and consent controls, and where the agent should propose rather than act.

Looker's LookML semantic layer is the reason it suits AI agents better than most BI tools: the definitions already exist, and an agent that queries through them inherits governed metrics and row-level security. This guide covers using LookML as the grounding layer, the Looker API for query execution, natural-language question handling, and keeping the agent inside the permissions users already have.

Power BI's semantic models hold the measures, relationships, and row-level security that make an agent's answers consistent with the reports people already trust. This guide covers querying semantic models through the execute queries API with DAX, running as the requesting user, grounding in model metadata, and returning answers with the measure, filters, and report link that let users verify.

Segment is the customer data pipeline that everything downstream depends on, which means an agent there can improve data quality across the whole stack or quietly corrupt it. This guide covers tracking plan governance, event schema and quality monitoring, profile and audience access, and the activation patterns that add value without introducing bad events into the stream.

Tableau's governance lives in published data sources and the workbooks built on them, and an agent that queries through those inherits certified definitions and permissions. This guide covers the REST API and VizQL Data Service for headless queries, using published data sources as the grounding layer, running as the requesting user, and returning answers with the workbook and field provenance users can check.

Amplitude holds the behavioural data product teams argue about, and an agent that answers their questions accurately changes how decisions get made. This guide covers API access and the event taxonomy, translating natural-language questions into governed queries, anomaly detection on key metrics, and keeping interpretation grounded so the agent explains the data rather than inventing a narrative.

Dropbox holds working files for teams that value simplicity over administration, which means an agent inherits both a clean API and folder structures nobody designed. This guide covers OAuth for individual and team accounts, reacting to file events, respecting shared folder permissions, and the organisation, extraction, and retrieval workflows worth building.

Gmail is the mailbox for Workspace organisations and a large share of small businesses, and its API is straightforward enough that the difficulty is entirely in doing it safely. This guide covers OAuth scope selection and app verification, Pub/Sub push notifications, label-based triage, drafting without sending, and the Workspace admin and compliance controls that govern the whole thing.

Google Drive is the working document store for organisations on Google Workspace, and its permission model is the most granular of the mainstream platforms. This guide covers OAuth versus domain-wide delegation, honouring Drive permissions in retrieval, handling native Docs, Sheets, and Slides, change detection, and the classification, extraction, and retrieval workflows worth building.

Twilio is the programmable layer beneath SMS, WhatsApp, and voice for a large share of customer communication, which makes it the natural place to put a conversational agent and the place where messaging regulation bites hardest. This guide covers channel differences, webhook-driven conversation state, consent and sender compliance, latency for voice, and escalation to people.

Meetings generate the decisions and commitments that never make it into any system, and Zoom holds the recordings and transcripts that could change that. This guide covers Zoom app types and OAuth scopes, accessing recordings and transcripts, the consent and notice obligations that govern recorded speech, and building summaries and action tracking people actually trust.

Outlook holds the highest-volume, most sensitive stream in the organisation, and Microsoft Graph gives an agent access to all of it with a single consent. This guide covers Graph permission models, mailbox and calendar access, the triage, drafting, and scheduling workflows worth building, and the controls that keep an agent from sending something it should not.

Box is where regulated enterprises keep the documents they cannot lose, which makes it a strong foundation for document agents and a place where permission mistakes are serious. This guide covers app authentication and as-user access, enforcing folder permissions in retrieval, using Box metadata, and the classification, extraction, and retrieval workflows worth building.