Transforming from SaaS to an Agentic Enterprise

TL;DR

• Two core customer pains: content compliance — the European Accessibility Act (EAA) in Europe and the ADA in North America both require large enterprises to keep content accessible — and content visibility, now that search has moved from traditional engines to AI-driven search.
• The answer is shifting left: detect and remediate issues at the moment developers, designers, and content creators produce the content, rather than scanning after publication.
• The delivery mechanism is an MCP server that plugs their agents into Lovable, Copilot, VS Code, and similar AI building tools so agents can call each other — collapsing integrations that used to take weeks or months into days or minutes.

Key Points

The problem: infinite content plus two kinds of compliance pressure (~01:21–01:23)

Siteimprove has been in the industry 20+ years with 5,500+ customers worldwide across the Global 2000 and Fortune 500 — financial institutions, healthcare, public sector, and more. Her framing: we now live in an AI-driven world with infinite content — web, mobile, social, documents — and large global enterprises are struggling with all of it.

The first challenge every customer raises is compliance, specifically accessibility. The European Accessibility Act (EAA) in Europe and the ADA (Americans with Disabilities Act) in North America both require large enterprises to keep their content compliant.

The second is content visibility: search has evolved past traditional search into AI-driven search — she asked for a show of hands and, at an agentic AI conference, essentially everyone was using AI search. Whether your content surfaces there is now a competitive question.

The third: in a world where everyone is a developer and everyone can generate content and designs, how do you ensure that content stays compliant and performs?

The response: shift left and an agentic content intelligence platform (~01:23–01:25)

Their answer is to shift left — address the problem at the time of creation, detecting and remediating issues as developers write code, designers produce designs, and creators generate content, instead of cleaning up afterward.

Last year they released an agentic content intelligence platform across four pillars:

1. Accessibility agents — detect and auto-remediate accessibility issues at creation time.
2. Conversational analytics agents — ask a question, get generated reports and dashboards.
3. Search agents — make content visible across AI-driven search surfaces.
4. Content strategy agents — content briefs, drafts, and descriptions.

Alongside that they released an MCP server enabling agent-to-agent integration across AI tools — Lovable, Copilot, VS Code. The benefit she emphasized is integration speed: what used to take weeks or months to bring to market can now be done in days or even minutes.

Demo: Lovable and Siteimprove agents calling each other (~01:25–01:27)

A recorded demo of the Lovable integration:

1. Ask Lovable which agents it has access to — it immediately detects the Siteimprove agents, Lovable agents, and browser agents.
2. Ask what the Siteimprove agents can do — it lists capabilities including detection, AI rules, and auto-remediation (exact product names in "To Verify").
3. Grab the URL of the site shown on the right and run a quick accessibility check — roughly five issues come back.
4. Prompt the Siteimprove agents and Lovable agents to work together: calls go through the MCP server, issues are identified, and remediation happens automatically.

Her conclusion: detecting and remediating these issues used to take weeks to months and now takes minutes. The goal is helping customers keep their infinite volume of content both compliant and performing — "it's not an either-or for us, it's an and, and that's all running on one single platform."

Quotes

"Our response … is to shift left — is to address this problem at the time of creation." (~01:23)

In a world where everyone is a developer, post-hoc scanning can't keep up with content creation.

"It's not an either-or solution for us. It's an and solution, and that's all running on one single platform." (~01:27)

Compliance and performance are both requirements, not a trade-off.