Unprompted: Context-Aware AI and What It Means for Marketing Teams

Welcome to Unprompted: The AI Marketing Brief, where I cut through the noise in AI news and research to show marketers what’s happening — and why it matters for your work, your team, and your career. 

I’m always singing the praises of AI. Whether you’re a seasoned marketer, a recent graduate just starting your journey, or someone without a technical background, I encourage you to dive into the AI pool headfirst. But here’s the catch: Research from the BBC shows that when AI is asked about current events, over half of its responses contain major errors. So before we jump into the exciting, innovative developments from companies like Google and Slack, let me take a moment on my soapbox: always double-check AI’s output. No matter how familiar you are with a tool — or how often you’ve used a particular prompt — verify the accuracy of its results.

Ok, now that’s out of the way, let’s get back to our regularly scheduled roundup. 

Groundbreaking BBC research shows issues with over half the answers from Artificial Intelligence (AI) assistants 

Website: BBC Media Centre 

Just the Facts: The BBC published research evaluating how well major AI assistants answer questions about current news, testing tools like ChatGPT, Microsoft’s Copilot, Google’s Gemini and Perplexity against 100 news prompts sourced from BBC content. Journalists found that over half (51%) of AI responses contained significant issues, including factual inaccuracies, misquotations and distortions of source material. Additionally, a meaningful share of answers that cited BBC articles introduced incorrect facts, and many responses had altered or fabricated quotes, raising concerns about AI reliability for news information. 

Why It Matters to Marketers: 

• With ~45 % of AI responses containing significant errors and ~31 % having serious sourcing problems, B2B teams using AI for market insight, competitive research or content fact-checking must build verification checkpoints into workflows rather than trusting outputs at face value to avoid embedding misinformation into strategy and executive briefs.  
• Because AI assistants confidently provide flawed answers — even on simple factual queries — marketing analytics and RevOps leaders need to treat AI-derived data as provisional, adding manual validation or second-level review before feeding into KPIs, forecasts or planning models.  
• The prevalence of errors, even in widely used AI tools, will force organizations to invest in critical evaluation skills, such as source assessment and context calibration, across teams. This will elevate accountability for how AI outputs are annotated, cited and presented in client or board communications.  
• As more buyers and stakeholders rely on AI assistants for information, persistent inaccuracies risk eroding trust not just in the tools but in brands that amplify those outputs. B2B leaders should consider governance frameworks that define acceptable AI use cases, monitor output quality and protect brand equity in external and internal messaging. 

Gemini introduces Personal Intelligence  

Author: Josh Woodward 

Website: Google Blog 

Just the Facts: Google has launched a beta feature called Personal Intelligence in its Gemini app that lets users connect apps such as Gmail, Photos, YouTube and Search to deliver highly contextual, personalized AI responses based on their own data. The feature works only if users opt in and choose which apps to connect, with privacy controls and transparency about how data is used. It’s rolling out first to Google AI Pro and AI Ultra subscribers in the U.S., with plans to expand more broadly over time. 

Why It Matters to Marketers: 

• Marketers can leverage Personal Intelligence to quickly synthesize insights from Gmail, Photos, YouTube and Search, reducing time spent gathering context and enabling faster, data-informed campaign decisions. The impact is immediate for teams managing content and audience research.
• By connecting AI to user-specific data, B2B teams could prototype more personalized messaging and account-based marketing experiences, signaling a shift toward hyper-contextual campaigns that respond to individual prospects’ behaviors — this is a medium-term strategic opportunity. 
• The opt-in, app-specific approach highlights privacy considerations; marketers must ensure their AI-driven workflows comply with consent and data protection rules, reinforcing accountability and ethical AI practices. This has an immediate operational impact. 
• Exposure to tools like Personal Intelligence underscores the need for AI fluency across marketing functions, from strategy to execution, enabling teams to experiment with AI-driven insights while understanding limitations. This offers longer-term capability building for the organization. 

Recursive Language Models (RLMs): The Clever Hack That Gives AI Infinite Memory 

Author: Corey Noles 

Website: The Neuron 

Just the Facts: Recursive Language Models (RLMs) are a new inference strategy developed by MIT researchers that lets AI systems handle vast amounts of context by treating large documents as an external environment the model can programmatically explore, instead of forcing all the text into the model’s limited attention window. RLMs work by loading data into a Python REPL and having the AI write code to search, partition and recursively call itself on relevant sections, stitching the results together into a coherent answer. This approach dramatically extends effective AI memory (theoretically to “infinite” context) while improving accuracy on complex reasoning tasks and keeping compute costs comparable to standard calls. 

Why It Matters to Marketers: 

• RLMs let AI systems process and reason over extremely large datasets — like multi-year customer logs, deep technical documentation or sprawling market research — without context limits, prompting B2B teams to rethink how generative AI can augment high-value insight workflows rather than just short-form drafts.  
• Because the model actively navigates and queries relevant parts of content instead of trying to remember everything at once, output accuracy and relevance on dense business problems (e.g., competitive analyses, legal/regulatory synthesis) rises, reducing hallucination risk in strategic decision support.  
• RLMs operate by programmatically interacting with external environments (e.g., REPL databases), meaning marketers and RevOps leaders should explore platforms or APIs that expose “context as database” paradigms, aligning martech stacks with tools that can scale beyond token-limited prompts.  
• As models evolve toward recursive, context-aware reasoning, traditional prompt-centric performance metrics will give way to benchmarks centered on reasoning depth, context-retrieval fidelity, and integration with enterprise knowledge graphs — pushing measurement frameworks in B2B analytics and governance to adapt. 

Meet the All‑New Slackbot: Your Personal AI Agent for Work  

Author: The team at Slack 

Website: Slack Blog 

Just the Facts: Slack has relaunched Slackbot as a context‑aware AI agent built natively into the platform that uses workspace data — messages, files, channels and permissions — to help employees synthesize information, draft content and act on work tasks without leaving Slack. The redesigned Slackbot delivers personalized, role‑specific assistance by understanding team context and workflow patterns, enabling more accurate, relevant output than generic AI tools. It respects enterprise security boundaries and only accesses information that users are permitted to see, with availability rolling out to paid Business+ and Enterprise+ customers. 

Why It Matters to Marketers: 

• By surfacing relevant messages, files, and channel activity within Slack, the AI agent can reduce context-switching, enabling marketing teams to execute campaigns and cross-functional projects faster, with immediate operational impact.
• Slackbot’s ability to draft summaries, messages and task prompts allows marketers to prototype campaign copy, internal updates and stakeholder communications more efficiently, improving day-to-day output quality.
• Operating within workspace permissions ensures sensitive marketing and customer data are protected, highlighting the importance of building AI workflows that maintain privacy and accountability. This is critical for enterprise B2B teams now.
• Embedding a context-aware AI agent into collaboration tools signals a shift toward AI-assisted team operations; marketing leaders should plan for long-term adoption strategies, AI literacy and integration into demand-gen and account-based workflows.