At our recent Auckland Brainy Breakfast event on 'Where AI is headed in 2026', we had so many questions come through at Q&A, we simply didn't have the time to answer them all. Luckily, two of our speakers have now answered all your questions in this Q&A article below. You asked, we delivered! A big thanks to Richard and Adnan for taking time to share their insights with us.
Brainy Breakfast Q&A follow-up with Richard Conway (The Optimisers) and Adnan Khan (Stitch)
Nearly 300 marketers gathered bright and early for the Marketing Association’s first Brainy Breakfast of 2026 - and the conversation quickly confirmed what many in the room already suspected: AI is no longer a side tool. It’s reshaping how brands are discovered, evaluated and chosen.
The focus of the session was where AI is headed in 2026 - from the rise of agentic AI to the implications of AI search, GEO (Generative Engine Optimisation), zero-click commerce and autonomous purchasing. Underpinning it all was a fundamental shift: customers are increasingly delegating research, comparison and even buying decisions to AI assistants.
If AI is building the shortlist, where does that leave your brand?
To explore that question more deeply, we asked Richard Conway, Founder of The Optimisers, and Adnan Khan, Co-founder of Stitch, to expand on some of the most pressing questions raised on the day.
Q1: What’s your specific advice for new start-ups maximising discoverability with AI agents (for investors and strategic partners)?
Richard recommends that you prioritise machine legibility from day one: structured data, clean feeds, and frictionless access to your pricing, availability, and policies so agents can ingest the facts quickly. Pair that with a deliberate rollout strategy for third-party endorsements, stories, and citations - startups make great news stories, so lean into media, podcasts, and local startup groups. The more citations, the better.
For investors/partners: be vigilant - many new businesses are essentially “wrappers” for AI agents. Look for founders with existing brand equity or connections that give “brand gravity,” and make sure someone on the investor side can distinguish genuine technical capability from “jazz hands.”
Q2: How are you prioritising optimising for agents now versus standard GEO given agentic conversions are still a small %?
Richard says that many fundamentals benefit both GEO now and agentic SEO later. In the short term, focus on being cited and summarised. But don’t wait for agentic commerce to become mainstream - the risk is becoming invisible. There’s no downside to preparing your technical and trust stack now; it’s a good business decision with likely future upside. (And in larger organisations, there’s often a lag between plan and execution - starting earlier matters.)
Q3: Data - “I’m told our data isn’t in a fit state so we can’t use AI. Is this the case?”
“Our data isn’t ready” has become the most expensive excuse in marketing according to Adnan Khan. Only 4% of organisations report fully AI-ready data, yet adoption is already 72%. If you wait for perfect data, you’ll never start - and AI is one of the best tools for improving imperfect data. Adnan’s view: assess readiness per use case, not across the whole business. Start with a high-value use case, get that data ready, deploy, learn, iterate. Many AI applications work well with 80–85% accuracy. Data readiness should be treated as a business outcome metric, not an IT purity test.
Richard echoes Adnan’s sentiment, pointing out that this is a common - and dangerous - excuse for inaction. You don’t need a perfect data lake; you need commercial liquidity. Start small: format pricing, availability, and core service SLAs into Schema.org (or publish an llms.txt) so your public commercial facts are structured and verifiable. AI doesn’t need your messy internal CRM to recommend you; it needs your outward-facing truth to be machine readable.
Q4: The elephant in the room - AI replacing human function. A retailer told their agency: “Cut 30% off your bill, use AI.”
According to Adnan, the 20–30% cost efficiencies are real, and deployment times can be cut dramatically - but “cut the bill” is reductive. Adnan points to a split in the market: some groups are cutting heavily, while others are investing and using AI to win growth. The roles being eliminated tend to be “middle execution layers” (data prep, basic drafting, reporting). The roles growing are strategy, orchestration, AI governance, and creative leadership. For NZ agencies, the existential issue is pricing: when tasks compress from hours to seconds, billing by the hour punishes efficiency. The shift from selling services to selling solutions is the big change.
Adnan points out that in 2026 an agency’s value isn’t just “using AI.” Humans are required to audit outputs for brand safety, compliance, and strategic nuance. You’re not paying only for the work; you’re paying for judgement and up-to-date thinking that prevents AI from hallucinating your brand’s reputation away. The best agencies win; the ones that lean on AI without care for accuracy, safety, and security are at risk.
Q5: How does AI redefine the customer path to purchase?
Adnan’s core point: the consideration phase collapses into an AI conversation that brands may never see. Discovery, evaluation, and shortlisting happen inside the tool before the buyer lands on your site - meaning marketing must increasingly influence the AI’s reasoning, not just human browsing. Structured, machine-readable product information becomes as important as creative, and “winning Google” doesn’t guarantee “winning AI citations.”
Richard adds that we’ve moved from a funnel to an adaptive loop.
The customer doesn’t “walk the path” anymore - they define the destination, and the agent builds the path in real time.
Q6: With zero-click commerce, how do performance marketing and performance platforms adapt?
Richard says that Google and Meta will likely shift from selling clicks to selling outcomes. In a zero-click world, spend moves toward in-platform conversions (buying inside ChatGPT or an Instagram agent). Performance marketers may shift from media buying to model influence - potentially paying to be the sponsored recommendation inside an agent’s reasoning process. Another angle: platforms could offer paid businesses special pricing, offers, or loyalty schemes that agents can surface preferentially. (Richard flags this as opinion - outcomes will be shaped by shareholder value and platform revenue models.)
Q7: How secure are these AI agents? (Cyber security)
Agentic systems don’t just chat - they plan, access data, call APIs, and take actions. Adnan points to emerging best practice: least privilege, human-in-the-loop for high-impact actions, validation against prompt injection, strong authentication for non-human identities, and comprehensive logging into security monitoring systems. The message isn’t “don’t use agents” - it’s “deploy them with the same rigour you’d apply to any system handling customer data.”
Richard says that this is more a question for IT/security teams, but one practical note: if your business isn’t using a secure handshake protocol (like MCP), agents may be blocked by corporate firewalls.
Q8: Case studies - what tools did companies use to implement agentic AI?
Adnan’s takeaway: there’s no single platform answer - enterprises are choosing tools that fit their ecosystem (custom builds, agent frameworks, CRM-linked agents, and automation stacks).
He highlights examples spanning insurance claims processing, airline concierge assistants, AI sales agents, and AI-optimised media buying — and flags n8n as a practical, lower-cost path for organisations without enterprise budgets, alongside frameworks like LangChain / LangGraph and other marketing-agent tooling.
Q9: Who wins when agentic AI marketing targets agentic AI businesses/consumers?
Adnan’s framing: interoperability standards and agent-to-agent protocols are emerging quickly, and agentic commerce could become enormous. When agents optimise on objective parameters (price, availability, reviews, returns, fulfilment reliability), emotional storytelling alone becomes insufficient. But humans still set the preference weights - so brand building remains essential to shape what consumers tell their agents to value.
The imperative is dual: emotional resonance for humans + machine readability for agents.
Richard points out that the winners are those brands people ask for by name, brands with clean data, and brands with strong third-party proof (reviews/citations). An agent doesn’t care about your homepage video - it cares about “validation mapping” (Reddit threads, G2 reviews, Wikipedia citations). The winner is the brand whose off-site proof is consistent enough to be a “zero-risk recommendation.”
The losers: commodity sellers with no brand preference, those relying only on on-site persuasion, and those with messy data and slow APIs.
Q10: What’s your opinion on OpenAI adding ads to ChatGPT, especially re: consumer trust?
Richard thinks this is a double-edged sword: it solves monetisation, but if users feel the assistant is basically a salesperson, they’ll leave. Brands that win will prioritise organic authority so they show up in the reasoning layer - not just in a sponsored slot.
Q11: What’s driving the shift toward agentic AI - is there proof consumers want it?
Adnan’s view: adoption is rising because shoppers get utility - faster research, more confidence, fewer returns, and increasing willingness to delegate parts of purchasing.
There are caveats: a meaningful share of consumers aren’t planning to use AI, and privacy concerns are high. But the trajectory is clear: growing use for research / comparison now, gradual movement toward autonomous purchase decisions next - with brands needing to deliver real utility, not gimmicks.
Richard adds that the shift is driven by search evolving into decision engines - people want life admin done for them. Richard points to consumer research indicating movement from assistive to agentic behaviour (and notes that practical delegation is a key driver).
Q12: How do you influence third-party sources mentioning your brand (Reddit, Wikipedia, etc.)?
Richard makes it clear that you can’t “SEO” Reddit or Wikipedia in the traditional sense - you must earn the mention. That means community participation and digital PR that creates high-authority citations. If AI sees consistent positive mentions on Reddit and citations on Wikipedia, it forms a “semantic cluster” of trust.
You influence the machine by influencing the humans the machine trusts.
Q13: Is there an ethical issue using AI to monitor customer messages for lead sourcing?
Adnan points to an important NZ-specific shift: the NZ Privacy Amendment Act 2025 introduces IPP 3A, taking effect 1 May 2026, requiring notification when personal data is collected indirectly (third parties, scraping, AI tools). He also references Privacy Commissioner guidance emphasising privacy impact assessments, leadership approval, human review, and te ao Māori considerations around data sovereignty.
The ethical path: prioritise first-party data, be transparent, use aggregated insights rather than surveillance, provide opt-outs, and keep human oversight.
Richard says it comes down to value vs intrusion. If monitoring helps provide a relevant solution at a moment of need, it’s concierge-like. If it’s harvesting data without consent, it’s dodgy. Transparency is the only fix: users must know when an agent is listening and why.
Q14: How does agentic AI impact high-cost, long sales-cycle B2B services?
In complex B2B, agents can remove friction in research and prep - improving conversion by focusing teams on high-intent accounts and reducing pre-sales effort. But Adnan stresses a critical counterbalance: in long-cycle, high-stakes B2B, AI should enhance human selling, not replace it. The goal is faster, smarter, more empathic, data-driven selling - augmentation.
Richard says, in B2B, agents handle the arduous middle: governance, claims, compliance, supplier shortlisting - comparing MSAs, checking security compliance, verifying case studies. A human may sign the final contract, but agents do the interim heavy lifting. If you aren’t “agent-ready,” you may be filtered out of the RFP before a human even sees your name.
Q15: If every brand is equal in the purchase process, how much more important is brand preference when you’re not in market?
According to Richard, brand preference becomes your insurance policy. If all else is equal, preference matters - and it must be built before the buyer is in-market so that when the moment arrives, your brand is the one their agent (and they) choose.
Q16: What strategy should brands take to surface meaningful differentiation - it’s one thing to be cited, how do we ensure we’re recommended?
Adnan draws a clear line: being cited and being recommended are different outcomes - and recommendations are where commercial value concentrates.
The practical playbook: audit AI visibility (most brands don’t), close “semantic completeness” gaps, invest in authoritative third-party placements (because your site is only a slice of what models cite), structure content for machine consumption (schema, clear entities, verifiable claims), and treat GEO as a distinct discipline from SEO. Early movers build compounding advantage.
Richard advises to give the machine a reason to choose you and it will. Don’t just claim “we’re faster” - publish verifiable attributes in structured data (e.g., specific latency metrics). AI agents don’t respond to fluff; they respond to measurable, checkable proof. If you make your differentiation machine-readable, you stop being a commodity.
Key Takeouts for New Zealand Marketers
Written by:
Richard Conway, CEO, The Optimisers
Adnan Khan, Managing Partner, Stitch NZ,