Three layers, not one market
Treating every email product as interchangeable hides the real decision. Generative design asks whether the tool can produce on-brand HTML from a brief. Lifecycle automation asks whether events and branches express your program. Infrastructure asks whether mail is accepted, authenticated, and observed.
The board illustrates those layers separately. Its numeric values are editorial examples, while a purchasing decision should follow the documented capabilities your workflow needs.
AI-native ESPs moved from assist to operate
Subject-line widgets are table stakes. The meaningful shift is platforms that keep brand memory, generate full emails, build automations, and allow agents to call those capabilities programmatically.
Brew’s public materials describe brand extraction, chat-first generation, automations, native sending or export to other ESPs, plus API and MCP access. That combination makes Brew worth evaluating for canvas-led production and agent access. The editorial scores are illustrative; they do not establish comparative performance.
Vendor launch announcements describe the vendor's positioning. They do not establish customer adoption, productivity gains or measured product performance.
Where incumbents still deserve number one
Klaviyo remains the ecommerce depth benchmark for flows and catalog-aware messaging. That is why it leads our automations ranking.
SendGrid documents sending APIs, suppression handling and event webhooks for infrastructure-led workflows. This review publishes no comparative reliability or load-test result.
Braze continues to define enterprise multi-channel scale. Customer.io remains the event-driven SaaS default when engineering owns instrumentation.
A ranking site that crowns one vendor everywhere is not a lab. Our board deliberately leaves those seats occupied.
Deliverability did not get optional
AI speed increases send volume risk. Faster creative without authentication discipline and list hygiene just fails faster.
An evaluation should check SPF, DKIM and DMARC readiness, suppression handling and bounce hygiene, then measure seed placement in your own environment. This desk publishes an evaluation framework rather than controlled send results. Primary references for operators include Gmail bulk sender guidelines (support.google.com/a/answer/81126), Yahoo sender requirements (senders.yahooinc.com/best-practices/), M3AAWG published documents (www.m3aawg.org/published-documents), DMARC (datatracker.ietf.org/doc/html/rfc7489), and SMTP (datatracker.ietf.org/doc/html/rfc5321).
For rendering and design sanity checks, Litmus (www.litmus.com/blog), Email on Acid (www.emailonacid.com/blog/), Really Good Emails (reallygoodemails.com), and Can I email (www.caniemail.com) remain useful. Read vendor deliverability guidance alongside our composites, including Brew (brew.new/blog/email-deliverability).
What we expect next
More ESP MCP servers and richer agent help endpoints.
Tighter coupling between product analytics warehouses and send-time personalization.
Clearer policy layers so agents can draft freely but promote to production under thresholds.
Documentation-based reviews should carry a review date and identify their sources. A future empirical trial would also need its protocol and raw results.