What VideoAmp and Rattle Actually Did
VideoAmp, an advertising-measurement company, laid off approximately 20% of its workforce - around 50 to 60 people, including its CTO. CEO Tony Fagan said the company is reorganising around agentic software development and repositioning itself as an AI-powered media-performance platform rather than another audience-measurement competitor.
Rattle's transformation is more extreme. The revenue-tech company cut headcount from roughly 70 to 15, rebuilt itself as AI-native sales platform Von, and recently reached about $1 million in annualised recurring revenue.
These are not isolated events. They are early evidence of a structural change in how martech and RevOps companies are built and operated.
The Architecture That Is Being Replaced
The traditional SaaS workflow architecture for marketing and revenue operations looks like this:
database - dashboard - human analyst - decision - execution
Each arrow in that chain represents a human handoff. Someone queries the database. Someone interprets the dashboard. Someone makes a decision based on the analysis. Someone executes the action. Someone measures the result.
The value of traditional SaaS was in making those handoffs more efficient: better interfaces, faster queries, cleaner visualisations, easier collaboration.
Agentic systems aim for a different architecture:
signal - reasoning - decision - execution - measurement - optimisation
The human handoffs are not made more efficient. They are removed.
That dramatically reduces the value of software whose primary function is moving information between humans.
The SaaS Disruption Is Not What You Think
The common framing is "AI will replace SaaS." That is imprecise and probably wrong for most categories.
CRM, analytics, attribution and ad-tech platforms are unlikely to disappear. The data they hold and the integrations they maintain are genuinely valuable.
The more accurate framing is: AI removes the human handoffs SaaS was built around.
The number of people required to query platforms, interpret them, move data between them, create reports, trigger routine actions and manage workflows can fall substantially. That is what VideoAmp and Rattle are demonstrating.
The future moat for martech companies becomes:
- Trusted data: Proprietary signals that agents need to make good decisions
- Integrations: Connections to the systems where actions need to happen
- Action authority: The permission to actually execute, not just recommend
Interface complexity - the traditional moat of enterprise SaaS - becomes a liability rather than an asset when agents are the primary users.
Which Marketing Roles Become More Valuable
The restructuring of martech workflows does not mean all marketing roles are at risk. It means the distribution of valuable skills changes.
Roles that become more valuable:
Workflow architects: People who understand how agent systems should be structured, what data they need, what actions they should be authorised to take, and how to measure their performance.
Commercial analysts: People who can evaluate whether agent-generated outputs meet commercial and strategic requirements - not just whether the data is correct, but whether the decision is right.
Governance leads: People responsible for ensuring agent systems operate within appropriate boundaries, comply with regulations, and maintain audit trails.
Agent directors: A new role emerging in forward-thinking organisations - people responsible for the portfolio of AI agents deployed across marketing and revenue operations, their performance, their interactions and their governance.
Roles under pressure:
Roles whose primary function is moving information between systems, creating routine reports, triggering standard workflows, or interpreting dashboards to make decisions that could be made algorithmically.
What This Means for B2B Marketing Teams
The VideoAmp and Rattle restructurings are early signals of a broader transition. B2B marketing teams should be asking:
Which parts of our workflow are human handoffs around software? These are the areas most likely to be automated first.
What data do we own that agents need? Proprietary first-party data, customer intelligence and commercial context are more valuable in an agentic world, not less.
How are we building agent authority? The question is not just whether to deploy agents but what actions they are authorised to take autonomously versus what requires human approval.
Are we measuring agent performance? Traditional SaaS metrics (seats, usage, feature adoption) do not capture agent value. Outcome-based measurement - pipeline generated, cost per qualified lead, revenue influenced - becomes the relevant metric.
Modi Elnadi is the founder of Integrated.Social, a B2B AI marketing agency in London specialising in agentic AI lead generation, AEO/GEO and performance marketing.
