Resources / For teams acting on the findings
Useful reading and tools
for the work ahead.
The studies that describe the same problems from other angles, and the tools that can take one recurring task off your team. What each one is, what it is good for, and where to start.
23 of 23 resources.
Tools you can use now · Open standard
Agent Skills, the open format for written AI methods
An open file format for writing down how a job should be done so that any compatible AI assistant follows the same instructions, with the reference material and scripts it needs kept alongside.
Cited in chapter 01 ↗Patterns and examples · Published design pattern
Andrej Karpathy’s LLM Wiki, a pattern for keeping company knowledge current
A published pattern for having an AI assistant build and maintain a set of linked notes from your source documents, updating them as facts change and flagging contradictions, instead of re-reading the raw files every time.
Cited in chapter 01 ↗Tools you can use now · Open-source project
Firecrawl’s AnyDoc, a document-to-text converter
A conversion tool that turns Word, PowerPoint, Excel, PDF and web files into clean plain text an AI assistant can read, with tables, headings and footnotes preserved.
Cited in chapter 01 ↗Products to watch · Open-source project
Y Combinator’s QM, a shared workspace for AI assistants
Open-source software for running one AI assistant across a company, in Slack and on the web, where each person and each channel gets its own memory, files, permissions and working area.
Cited in chapter 01 ↗Patterns and examples · Vendor-published case study
Anthropic, how Zapier connects AI drafts to Docs and Slack review
A vendor case study describing how Zapier, a remote company the story puts at about 360 people, put AI assistants into everyday work, including a marketing loop where drafts land in Google Docs and a Slack message asks the team to review.
Cited in chapter 03 ↗Published research · Research summary
Gartner, marketing technology utilization research
Gartner’s annual survey of marketing technology leaders, whose 2025 edition reports that marketing teams use 49% of the technology capability they have bought, and that nearly half of the teams piloting vendor-supplied AI agents say they fall short of what was promised.
Cited in chapter 02 ↗Tools you can use now · Product documentation
OpenRouter, routing requests across AI providers
A paid gateway that lets a company call hundreds of AI models from many vendors through one account and one connection, with rules for which vendor each request may go to and a backup if the first one fails.
Cited in chapter 02 ↗Tools you can use now · Vendor product description
Treg, pay-per-task access to specialist marketing tools
A pay-per-call gateway that gives an AI assistant one login to dozens of marketing data and publishing tools, from search rankings and backlinks to contact details and social posting, priced by the call rather than by the seat.
Cited in chapter 02 ↗Published research · Industry analysis
Chiefmartec, State of Martech 2026
The annual State of Martech report, whose 2026 edition finds the martech landscape has stopped growing, that thousands of applications can now be operated by AI assistants directly, and that marketing roles are shifting from running tools to deciding what assistants need to know.
Cited in chapter 02 ↗Published research · Survey
Capgemini, CMO playbook 2025
Capgemini’s third CMO Playbook, a survey of 1,500 marketing executives at companies with $1B or more in revenue across 15 countries, on how marketing is funded, governed and using AI.
Cited in chapter 03 ↗Tools you can use now · Product documentation
Microsoft, Copilot administrator controls
Microsoft’s administrator documentation for deciding whether Copilot may use Anthropic’s Claude models, the setting that decides which model an employee’s Copilot in Word, Excel and PowerPoint is using.
Background reading ↗Products to watch · Product documentation
xAI’s Grok Bot, an assistant with its own computer and tools
A paid service that gives a person named AI “teammates” running on a hosted cloud computer, each with a job description and a memory, that carry multi-step tasks across websites and applications and ask for approval before risky steps.
Cited in chapter 03 ↗Products to watch · Product design account
Muse, an assistant that keeps context and asks before it acts
Meta’s personal AI agent, launched in the US in September 2026, that works on tasks in the background on its own secure cloud computer, keeps a memory, and asks before any purchase or action with consequences.
Cited in chapter 03 ↗Patterns and examples · Engineering account
Anthropic, engineering notes on long-running agents
An engineering note on how to make an AI assistant keep making progress on a large software project across many sessions, when each session starts with no memory of the last, using a checklist file, a progress log and a fixed start-of-session ritual.
Cited in chapter 03 ↗Products to watch · Open-source project
Block’s Buzz, a shared workspace for people and AI agents
An open-source chat workspace, self-hosted on infrastructure you own, where people and AI agents sit in the same channels and every message, change and approval is a signed, searchable record.
Cited in chapter 04 ↗Tools you can use now · Product documentation
Canva, design tools that AI assistants can use
Canva’s connector that lets an AI assistant create, edit, find, resize, comment on and export Canva designs from a chat, working from your brand templates where your plan includes them.
Cited in chapter 04 ↗Published research · Survey
Marketing AI Institute, State of Marketing AI 2025
A survey of 1,882 marketers drawn from the institute’s own audience, people already interested in AI, on how far their teams have got, what stops them and who owns it.
Background reading ↗Published research · Survey
Jasper, State of AI in Marketing 2026
A survey of 1,400 marketers, fielded in late 2025, on AI adoption, the ability to prove return, what blocks scaling and who owns AI workflows.
Cited in chapter 04 ↗Published research · Qualitative study
Callan Consulting, State of AI in Technology Marketing 2026
Interviews with 19 CMOs and marketing leaders, mainly in B2B technology, conducted in late 2025 and early 2026, on how AI has changed their departments and what none of them can yet measure.
Cited on the homepage ↗Published research · Survey
Norwest, B2B benchmark 2025
A benchmark of 177 sales and marketing leaders at venture- and private-equity-backed B2B companies, fielded in mid-2025, on budgets, performance measures and how AI actually got into use.
Cited in chapter 01, 04 ↗Published research · Survey
Anteriad, B2B Marketing Edge 2025
A survey of 466 senior B2B marketers at companies with 250 or more employees, fielded in early 2025, linking confidence in data to speed and growth in the marketers’ own accounts.
Background reading ↗Published research · Survey
Salesforce, State of Sales, seventh edition
The seventh edition of Salesforce’s sales survey, 4,050 sales professionals in 22 countries surveyed in late 2025, on AI agents in sales, tool sprawl and what buyers now expect.
Background reading ↗Published research · Survey and analysis
Microsoft, Work Trend Index 2025
Microsoft’s 2025 Work Trend Index, a survey of 31,000 knowledge workers in 31 markets plus Microsoft 365 usage data, on how leaders plan to use AI agents and how far their employees agree.
Background reading ↗