Media Agency Of The Year: MRM's AI Relationship Management
We've entered the agentic epoch. AI agents now sit between people and brands, changing how search, discovery, content, and commerce work. Marketing isn't just B2C or B2B anymore - it's agent-to-agent. That shift is already hitting revenue, operations, and the way teams make decisions.
MRM's move to AI Relationship Management (ARM) meets that shift head-on. As MRM Global CEO Grant Theron put it, "because in an AI-mediated world, how can you have a solid, valuable, long-term relationship between a consumer and a brand?" The team treated ARM as an engineering problem: organize data, define interfaces, and build the pipes agents can trust. The practice launched in September 2025 with data readiness at the core.
CRM has long been a ~$100B category built on identity, consent, and lifetime value. But AI agents now mediate intent and choices, so the most human-centric part of marketing is the most exposed. One client saw a 30% drop in commerce traffic as LLMs rerouted consumer behavior. The channel didn't vanish - it moved to agents.
Standards are arriving to normalize this new exchange. IEEE's P7012 enables machine-readable data transfers between agents representing consumers and brands, with Project VRM branding it "MyTerms." Consumers set the terms; brand agents respond. See IEEE P7012 and Project VRM's overview.
This promises to reduce information asymmetry by making value exchange explicit and bidirectional. Data becomes the proxy for trust, preference, and sustainability of the relationship. In that context, MRM's ARM mantra is blunt and useful: "Your customers trust AI. Make AI trust your brand."
What this means for management
Treat agents as a primary channel, not an experiment. Update goals, budgets, and org design accordingly. Here's a pragmatic starting point.
- Audit exposure: quantify traffic and revenue influenced by AI assistants, LLM browsers, and shopping agents. Track agent-originated referrals separately.
- Data readiness: clean identity graphs, consent states, and product catalogs. Add machine-readable policies, pricing, and availability.
- Agent interfaces: publish reliable endpoints (APIs, feeds, embeddings) that agents can query. Provide structured responses, citations, and provenance.
- Brand policy for machines: define guardrails, safety constraints, refund rules, and service levels in machine-readable form. Keep them current.
- Measurement: add metrics like agent satisfaction score, terms match rate, opt-in acceptance rate, agent referral conversion, and cost per resolved intent.
- Compliance by design: prepare for MyTerms-style negotiations. Automate consent logging, data minimization, and retention controls.
- Content for agents: ship structured specs, FAQs, troubleshooting steps, and comparisons. Make it easy for agents to recommend you without hallucination.
- Operating model: appoint an ARM lead, set up an agent red team, and create SLAs for agent-facing APIs and knowledge updates.
- Partnerships: work with marketplaces, device ecosystems, and assistant platforms where your customers spend time.
- Budgeting: reallocate a slice of search/social spend to ARM infrastructure, data quality, and agent performance testing.
Guardrails that build machine trust
- Be explicit: publish terms, warranties, and constraints in structured formats. Reduce ambiguity.
- Be verifiable: include citations, timestamps, and source IDs for all claims. Prefer signed artifacts.
- Be safe: test prompts, jailbreak resistance, and misuse cases. Log and fix failure modes fast.
- Be reciprocal: respect consumer MyTerms. Offer clear value for data and choices, with easy opt-outs.
- Be minimal: collect only what's needed for each intent. Prove it with data path diagrams.
90-day execution plan
- Days 0-30: Baseline agent-influenced traffic and revenue. Map top intents and questions. Add schema and structured data to priority pages.
- Days 31-60: Stand up agent endpoints for pricing, availability, returns, and support. Build a lightweight brand policy file for machines.
- Days 61-90: Pilot a MyTerms-compatible flow with an opt-in segment. Track terms acceptance, satisfaction, and conversion. Iterate weekly.
Upskill your team on agent operations, evaluation, and prompt policy. If you need a curated path for managers and marketing leads, explore these resources: AI courses by job.
The center of gravity just moved from your website to the agent layer. Brands that design for agents - with clear terms, verifiable data, and fast interfaces - will compound advantage. The rest will pay higher acquisition costs to keep up.
"Your customers trust AI. Make AI trust your brand."
Your membership also unlocks:
