Automationhigh impact

OpenAI Builds Agents for Everything as Forrester Warns Most AI Projects Stall

By InnovaAI Research2 min readTechcrunch

OpenAI is expanding its AI agent ecosystem to cover virtually every business task, while Forrester research finds that most organizations are generating growing lists of AI pilots and proofs of concept without converting them into meaningful business outcomes. For marketing agencies, the gap between experimentation and real results is now the defining challenge.

Key Facts

01OpenAI is expanding its AI agent ecosystem to cover virtually every business function, per a TechCrunch report dated August 24, 2026.
02Forrester finds that organizations are accumulating AI use cases and pilots without converting them into meaningful business outcomes.
03The core problem is not a shortage of AI capability but a shortage of focus and measurement discipline.
04Adding more AI tools without auditing existing ones accelerates the pilot accumulation problem.
05A 30-day measurement window tied to a specific client metric is a practical filter for deciding which AI workflows to keep.

Why does this matter for agencies?

▶The expansion of OpenAI's agent portfolio will increase pressure on agencies to adopt new tools before current ones are proven, compounding the problem Forrester identifies.
▶Clients are becoming more sophisticated about asking what AI investments are actually delivering, making internal documentation of outcomes a business development asset.
▶Agencies that establish a structured review process now will spend less on underperforming tools and build a more credible AI story for prospects.
▶Without a selection filter, a growing agent ecosystem becomes a cost center rather than a competitive advantage.

What should agencies do?

Audit every AI tool and automation currently in your stack: for each one, name the metric it was meant to improve and confirm whether you have measured it in the last 30 days.

medium effort

Establish a written internal policy requiring every new AI pilot to have a named owner, a defined deliverable, and a 60-day review date before it is added to the stack.

low effort

Connect at least one existing AI-assisted workflow to a client-facing reporting dashboard in AgencyAnalytics, Databox, or Whatagraph so the outcome is visible to both the client and your internal team.

medium effort