# Inference > An independent marketing studio for the AI rebuild. Publication, products, and consulting from Thom Cummings — strategy and systems for teams serious about scaling what actually works. Base URL: https://thisisinference.com/ Founder: Thom Cummings (Brighton, UK) Contact: thom@thisisinference.com --- ## Home — https://thisisinference.com/ ### Hero # inference. *build smarter marketing systems* ### Substack CTA Follow along with weekly annotations. (Substack subscribe form: https://annotationsbyinference.substack.com) ### The thesis — Marketing is being rebuilt as a system that needs design. AI didn't add a layer to marketing. It's rebuilding the system underneath it. Strategy, team, tools, and workflows used to be separate decisions; now they shape each other, and AI accelerates every weak point between them. Faster isn't better — it just scales whatever you already had. Good strategy scales beautifully. Bad strategy scales — and compounds. Most teams are bolting Claude onto a stack that was never designed for it. Inference is an independent marketing studio for that rebuild, at the intersection of strategy and systems. Craft and speed, combined with pattern matching you can't prompt for. We help companies build marketing systems that are worth scaling — positioning, growth architecture, operational infrastructure — so when AI accelerates everything, it's accelerating something good. Built on 20 years of building marketing engines at some of Europe's fastest-growing startups — SoundCloud, Grover, Pitch, Prolific. Now collaborating with teams rebuilding their strategy and GTM into systems that hold up to what's coming next. — Thom Cummings, Founder, Inference ### The publication — annotations Weekly guides, tool breakdowns, and field notes from the frontline of implementing smarter marketing systems. - Weekly writing on AI-native marketing in practice - Guides and tool breakdowns you can actually use - Paid subscribers unlock the Blueprints library → Read: https://thisisinference.com/annotations ### The products — blueprints Standalone marketing systems that rebuild each layer of an AI-native stack. Use them one at a time, or combine them into a full system. - Real, working systems — not templates or playbooks - Multi-layered workflows tested in live environments - Buy individually, or subscribe on Substack for the full suite → Browse: https://thisisinference.com/blueprints ### The consulting — collaborations For when your marketing ambitions have outgrown what's underneath them. Hands-on engagements that install the system that rebuilds your marketing. - Strategic Marketing Partner — senior thinking, in the work with you - Fractional CMO — embedded marketing leadership without the full-time hire - Systems Build — install your marketing system for brand, growth, agent teams → Work together: https://thisisinference.com/collabs ### Trusted by Prolific · Pitch · Grover · SoundCloud · Radicle · CareerFoundry · Betfair ### CTA — let's talk. We'll dig into where your marketing stands, where the gaps are, and how Inference can help. → Book a call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## About — https://thisisinference.com/about # Hi, I'm Thom Cummings I've spent 20+ years building and leading marketing at tech companies — from early stage through to scale. SoundCloud, Grover, Pitch, and most recently Prolific, driving 10-100x growth and building world-class brands. Inference is an independent marketing studio. It sits at the intersection of marketing strategy, systems thinking, and AI — the three things I believe will define the next era of how companies grow. I'm the principal. I lead every engagement personally: the strategy, the key sessions, the decisions. Where a brief needs specialist capability, research, design or technical build, trusted collaborators join in, named and with real roles. You always know who is doing the work. I write about what I'm learning as I rebuild my own studio with AI (annotations), build marketing systems as products (blueprints), and work directly with companies navigating this shift (collaborations). Based in Brighton, UK. ### Say hello. Always happy to meet people thinking about marketing, AI, and systems. - Email: thom@thisisinference.com - LinkedIn: https://linkedin.com/in/thomcummings --- ## Annotations — https://thisisinference.com/annotations # annotations *The publication.* Weekly guides, tool breakdowns, and field notes from the frontline of implementing smarter marketing systems. Subscribe on Substack for exclusive writing and access to the Blueprints library. ### Featured writing **A central hub for AI-native teams** — *April 2026, coming soon* Teams using AI are often dealing with fragmentation, context drift, and version control issues. The missing piece of the puzzle is a centralised context hub, powered by a private MCP server. This is the guide to the latest Inference Blueprint. **Positioning is a system, not a statement** — *April 2026, coming soon* Positioning work used to be a workshop and a set of slides. Often it stopped there. Here's what it looks like when you treat it as living infrastructure that actually influences product, sales, and marketing. **AI didn't change marketing. It exposed it.** — *April 2026, coming soon* The gap between teams that had real strategy and those that had volume dressed up as strategy? AI made that visible overnight. ### Don't miss one. New annotations land every week. → Subscribe: https://annotationsbyinference.substack.com --- ## Blueprints — https://thisisinference.com/blueprints # blueprints *The products.* Each blueprint is a standalone marketing system rebuilding a specific layer of an AI-native stack. They work as individual modules or combine effectively as a complete system. Available to buy one-off or subscribe to my Substack for access to the full suite. ### Available systems **01. Marketing Audit** — *Coming soon* A full-system assessment of your marketing maturity across 6 broad categories and over 25 sub-topics — from brand and positioning through to demand generation, content, analytics, and team capability. You'll get a scored breakdown of where you are today, a clear view of your biggest gaps, and a prioritised roadmap for what to fix first. Built so you can re-run it quarterly and track progress over time. → Try the free four-minute Marketing Maturity Quiz, which samples from the same framework: https://thisisinference.com/blueprints/audit/quiz **02. Build Your Team's Context Layer** — *Coming soon* The structured knowledge base that makes AI tools actually understand your business — positioning, audience, messaging, competitive intel — captured as markdown your whole team and every agent can draw from. The substrate to build before anything else in your AI stack. **03. Build Your First Agent Team** — *Coming soon* A working marketing agent team, not another explainer. Learn what's inside an agent, how teams compose, and walk away with a downloadable org you can run — Strategist, Writer, Editor, Distribution, SEO — designed the way you'd hire one. **04. Custom MCP Server** — *Coming soon* A bespoke MCP server that distributes your team's skills, context, and accumulated knowledge through Claude's connector UI. Your operating system, queryable by every agent on your team. **05. Skill Forge** — *Coming soon* The production system behind every Inference skill. A curated pipeline — taxonomy, researcher agent, opinionated synthesis, evals — for building best-in-class marketing skills that beat the generic alternatives. **06. Positioning Stack** — *Coming soon* A four-skill bundle covering the full strategy-to-execution chain: positioning, brand architecture, messaging framework, language-market fit. Anchored in Dunford, Ogilvy, Jackson, Raskin, and Lerner. Use one layer or the whole pipeline. **07. Brand Strategy** — *Coming soon* Brand Inference — an agency-grade brand strategy process run entirely inside Claude. Immersion research, voice-AI stakeholder interviews, strategic synthesis, two-to-three strategy directions, and a downstream messaging cascade. Without the agency fees. **08. Content Operating System** — *Coming soon* A modular content pipeline for high-quality, on-voice output across channels. Audience profiles, platform-specific writing styles, lookalike pattern mining, a draft enricher, and a monthly feedback loop that improves the skills themselves. **Bespoke — Need something different?** Like what you see but need something tailored to your specific challenge? Let's build it together. → https://thisisinference.com/collabs ### Get early access. Be the first to know when new blueprints ship. Subscribe for updates: https://annotationsbyinference.substack.com ### Frequently asked. **What exactly is a blueprint?** A blueprint is a module in a marketing system. It's not a template, not a playbook, but a real, working example that you can plug into your stack. It codifies best practice marketing strategy alongside AI-native workflows. **Do I need technical skills to use them?** Yes and no. Blueprints are designed for marketing leaders and operators and are built to be plug and play. However they require some knowledge of the AI-native marketing stack (Claude Code, Github, MCPs) but are easily learnable and come with implementation guides. **Why are the blueprints not free?** There are lots of skills and plugins available for free on Github or LinkedIn, true. Many are great and have become indispensable parts of my stack. Blueprints are a little different — they are complex, multi-layered workflows (almost mini apps) that have had hundreds of hours build time and, crucially, already tested in real world situations. They're available to buy individually but the best way to get them is to subscribe to my Substack. **When will the first blueprints be available?** Soon. The Marketing Audit and Custom MCP servers are in final polish mode. **Can I use blueprints alongside a collaboration engagement?** Absolutely. In fact, that's often the best way. Blueprints give your team the system; a collaboration engagement gives you the strategic guidance to implement it. --- ## Collaborations — https://thisisinference.com/collabs # collaborations *The consulting.* For when your marketing ambitions have outgrown what's underneath them. Inference collaborations install the system underneath it — across strategy, team, tools, workflows, and AI. For Series A to growth-stage B2B tech teams where marketing now needs more structure, leverage, and scale. ### Two ways to work together. Partner with me for senior judgement on tap — or let's build your marketing system together, sprint by sprint. **Partner with me** **Strategic Marketing Partner** — *Rolling monthly · Founders & heads of marketing* A senior marketing brain in the work with you — weekly sparring, second opinions, and pattern matching you can't prompt for. → Detail: https://thisisinference.com/collabs/strategic-marketing-partner **Fractional CMO** — *3-6 month engagements · 2-3 days/week · For founders* Embedded marketing leadership without the full-time hire — I lead the function, own the outcomes, and coach the team. → Detail: https://thisisinference.com/collabs/fractional-cmo **Systems Build — upgrade your marketing system** Six modular sprints. Run one, or sequence several into a full build. **01. GTM Strategy** — *3-4 weeks · the front door* Decide what marketing should actually be doing next, and commit it to a 90-day plan with owners, budget and one metric each — plus an explicit register of what you're refusing this quarter. Underpinned by a full marketing maturity assessment across six areas and 29 sub-areas, scored against stage-appropriate targets. Three live sessions; the middle one is where you argue with the diagnosis before the plan is written. → Detail: https://thisisinference.com/collabs/sprints/gtm-strategy **02. Brand Foundations** Positioning, messaging and brand — installed as the foundational layer everything downstream draws from. → Detail: https://thisisinference.com/collabs/sprints/brand **03. Website Growth Engine** A website that works as a growth engine, not a brochure — brand and positioning improved, conversion and automation built in, then kept optimising. → Detail: https://thisisinference.com/collabs/growth-engine **04. AI Readiness** — *Opens with a workshop* Map where AI and agents change the economics of your marketing. You leave with a sequenced opportunity map and the first builds scoped. → Detail: https://thisisinference.com/collabs/ai-readiness **05. Context Layer Infra** — *Coming soon* The substrate under your AI stack — a structured context layer, a custom MCP server, and a Claude-ready setup every agent draws from. **06. Agent Teams** A complete marketing team, designed and deployed — agents in the right roles, humans where they count. → Detail: https://thisisinference.com/agent-teams ### Frequently asked. **What does fractional CMO actually mean?** It means you get a senior marketing leader embedded in your team part-time. I attend standups, run strategy, and own outcomes — without the overhead of a full-time executive hire. **Can I run just one sprint, or do I need the whole system?** Either. Each Systems Build sprint is self-contained, so you can start with the one where your marketing's weakest. Most teams run one, see the value, then sequence the next. There are two usual front doors: GTM Strategy, if the open question is the broader one — who you're for, how you get to market, and where growth actually comes from. Or AI Readiness, if you already know you want an AI adoption programme and need to map what to build and in what order. **What kind of companies do you work with?** Mostly B2B tech companies — from Series A through to growth stage. I've spent 20 years in the space across companies like SoundCloud, Pitch, Grover, and Prolific, so that's where I do my best work. **Who actually does the work?** I lead every engagement personally: the strategy, the key sessions, the decisions. Where the work needs specialist capability, trusted collaborators join with real, client-visible roles. You get one accountable lead, from proposal to handover. **What's the minimum engagement length?** The Strategic Marketing Partner runs on a rolling monthly basis. Systems Build sprints tend to run 2–6 weeks each. Fractional CMO engagements start at 3 months. Pricing is shared on a fit call. **How do I know which option is right for me?** Book a call and we'll figure it out together. Most conversations make it obvious within 15 minutes — just an honest assessment of whether and how I can help. ### Start with a conversation. We'll dig into where your marketing stands, where the gaps are, and how Inference can help. → Book a call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Strategic Marketing Partner — https://thisisinference.com/collabs/strategic-marketing-partner # Sparring partner on demand. *Collabs — strategic marketing partner.* You get senior judgement without the cost of a senior hire, applied to the strategic decisions you're making right now. Sparring on direction, second opinions on work in flight, pattern matching you can't prompt for. ### What you get - **Weekly 45-min sparring session** — a standing slot to surface what's on your plate, turn vague worries into specific next moves, and pressure-test the calls you're about to make. - **Async access between calls** — quick second opinions on docs, decks, hires, and decisions, without booking a meeting. - **A senior brain at your side, every week** — light-touch, high-leverage. Designed so you reach for it when it matters. ### Who it's for - Founders making senior marketing calls without a senior marketer in the room. - Heads of marketing in scale-ups who want coaching and an objective sparring partner alongside the day job. - Teams who already have execution covered but need strategic pattern matching. ### How it's different from fractional CMO The Strategic Marketing Partner is a senior partner in the work with you — sparring partner, second opinions, pattern matching, async input. Fractional CMO is embedded leadership — multiple days a week inside the team, owning the strategy and the outputs. It fits when you have a marketing team you want to complement with senior thinking alongside. Fractional CMO fits when you need the leader. → Book a call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Fractional CMO — https://thisisinference.com/collabs/fractional-cmo # Fractional CMO for teams rebuilding marketing for the AI era. *Collabs — fractional CMO.* You need a CMO. Just not every day. Senior marketing leadership embedded in your team — with the judgement, ownership, and operating experience of a full-time hire, plus the AI-native perspective to redesign how the function works. ### Embedded leadership. Owned outcomes. No long-term lock-in. Fractional CMO is the senior marketing leader inside your team — accountable for the strategy, leading the function, owning the outcomes. Different in kind from an advisor (who spars) and from an agency (who executes). - **Embedded** — inside the team, not outside the wall. Joining standups, leadership meetings, key Slack channels. Treated as the marketing leader, not a vendor. - **Accountable** — owning outcomes, not just recommendations. Strategy, hiring, vendor management, in-quarter execution oversight. - **AI-native** — most fractional CMOs are operating from a pre-AI playbook. I bring both the operator pattern matching from the old playbook and the AI-native experience of rebuilding for the new one. ### Shape of an engagement - **2-3 days per week** inside your team. - **3-6 month engagements** — heavier upfront investment, sustained momentum through the middle, deliberate handover at the end. - **Team coaching built in** — the goal is to bring your team to a place where you don't need me, through systems, clear planning, and coaching. ### When it's the right fit - You're seed to Series B and need senior marketing leadership without a full-time hire. - You have (or want to build) a marketing team and need someone to lead them. - You want strategy and execution oversight, not just recommendations. ### When something else might fit better - You already have a senior marketing leader — look at the Strategic Marketing Partner for sparring rather than embedding. - You want pure hands-on execution — a fractional CMO leads, you'll still need executors below. → Book a call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Website Growth Engine — https://thisisinference.com/collabs/growth-engine # Your marketing site, engineered to grow. *Collabs — website growth engine.* Your website is your biggest marketing asset — and for most teams the most under-optimised. The average B2B website converts just 2–5% of its visitors; the rest arrive and leave. The Website Growth Engine rebuilds it as a system: brand and positioning improved, conversion designed in, and automation doing the compounding. Then it keeps optimising, rather than going stale. ### What you get - **A brilliant website** — brand and positioning brought through. A website you can be proud of, built by someone who understands marketing, design, and growth. - **A system for growth** — positioning-led architecture and deliberate conversion paths, so traffic turns into pipeline instead of leaking away. - **Automation built in** — a starter pack of optimisation routines (CRO, SEO, AEO) automated into the site, so it keeps improving without a person babysitting it. ### Why it's an engine Most websites are finished the day they ship; this one is just getting started. With the Website Growth Engine, the site is a living system, not a static artefact — it keeps updating, refreshing and optimising itself, and every improvement is captured as a skill, automation or routine so the gains run on autopilot. - **Always current** — the copy, pages, structure and performance keep improving continuously, so the site compounds instead of decaying. - **Improvements on autopilot** — each optimisation becomes a reusable skill or routine (CRO, SEO, AEO, conversion work) that runs itself and stacks up over time. - **Your team, freed up** — with maintenance and optimisation automated away, the team spends its time on higher-leverage work only people can do. ### The offer The whole system in a single scope — no phased upsell. The build is everything; the retainer afterwards is optional. - **The Build (the core)** — a complete website rebuilt as a marketing system. Everything is included: - A complete custom website — every page your story needs (home, product/solution, about, contact, campaigns); no fixed page cap. - Brand and positioning brought through every page. - Conversion engineered into the structure of every page. - A starter pack of automated optimisation routines — CRO, SEO, AEO. - Answer-engine ready — structured so ChatGPT, Claude and Perplexity can find and quote you. - A reusable component library your team owns and builds on. - Set up, configured and handed over live — yours to run. - **Positioning add-on** — a compressed positioning pass bolted onto the build, for teams whose positioning isn't razor-sharp yet. Locked before the build starts. - **The Retainer (the ongoing engine)** — hands-on optimisation every month: tweaks, new pages and changes led by what the data shows, new automations built as opportunities emerge, and growth experiments run and read together — with a human calling which move is worth making next. Ongoing and uncapped. For existing Website Growth Engine clients. ### Who it's for - Professional services firms who need a credible online presence built by someone who understands marketing, not just web design. - Startups with clear positioning who need a site that works as a growth asset. - Fractional executives and solo operators who want a system, not just pages. - Teams who've outgrown their current site and need the next version built right. ### Frequently asked **I already have a website. Why would I replatform?** You wouldn't — unless the one you have isn't working as a growth asset. If traffic lands and nothing happens, or your positioning has evolved and the site hasn't, the rebuild pays for itself in the conversions it captures. **How does the build work?** The build runs on Ploy, which compresses the slow, mechanical part of a website build so the time goes on strategy and craft instead. You get a component library your team can build on, and your site stays visible to AI search engines — something most AI-built site platforms fail at. Inference is a Ploy agency partner. **What if my positioning isn't clear yet?** Add a positioning pass to the build — your positioning gets locked before it starts. If the gap is larger, a Foundation Sprint comes first, and the build follows once the strategy is solid. **Do I need my own Ploy account?** Yes — you set up a Ploy Pro workspace and add Inference as a collaborator to build in it. That workspace, and the site inside it, stay yours after handoff. The Ploy subscription is billed by Ploy and sits separate from the build. → Book a pre-build call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Marketing System Map — https://thisisinference.com/marketing-system # An interactive map of the whole marketing system. *Marketing system map.* An interactive visualisation of the Inference Marketing System Map — see how familiar marketing strategy, research, campaigns, measurement, technology and judgment connect into an AI-native system that learns from every cycle of work. The map is an interactive web app: explore the components and their relationships at the URL above. Its companion self-assessment is the Marketing System Diagnostic. --- ## Marketing System Diagnostic — https://thisisinference.com/marketing-system/diagnostic # A ten-minute self-reported diagnostic across the seven components of the Inference Marketing System Map. *Marketing system diagnostic.* A ten-minute, self-reported diagnostic across the seven components of the Inference Marketing System Map. No email or account required. Your answers stay in your browser and are used to create the downloadable result — no data is sent to Inference. The diagnostic is an interactive web app: open it at the URL above. --- ## Brand Foundations Sprint — https://thisisinference.com/collabs/sprints/brand # Refresh your brand foundations. *Collabs — brand foundations.* Architecture, positioning, essence, and messaging — the four decisions everything else follows from. A focused sprint to lock these in, for B2B SaaS teams whose brand strategy needs to catch up before the next phase of growth makes the gaps expensive. ### Why now Brand foundations are worth refreshing when the business moves. A new stage (moving up-market or into a new segment), a second product (deciding how the brand and products relate), or simply enough time passing (strategy that fit two years ago has quietly drifted) — each is a natural point to make sure the brand strategy still fits where you're heading. ### The Foundation Model Every Brand Foundations Sprint runs through the same four-layer model. Architecture at the base, messaging on the surface — each one builds on the layer below, and each one ships twice: as a document your team reads, and as structured context your team and your tools both run on. - **Layer 01 — Architecture.** How the brand and products relate. The structural decision that underwrites positioning, messaging, and product clarity — the hardest layer to reverse, which is why it's set first. *Output: a documented architecture decision, with the rationale.* - **Layer 02 — Positioning.** Where you sit, and why you win — the competitive frame and the customer you're for. *Output: a positioning system — statement, frame, value props, proof points.* - **Layer 03 — Essence.** The few words at the centre — the core idea and character that make the brand feel like yours. *Output: brand essence, personality, and the signature line.* - **Layer 04 — Messaging.** The words your team uses — strategy turned into language for sales decks, the site, and hiring conversations. *Output: a messaging house — boilerplate, value props, talk tracks.* Two capstones sit beside the four layers: - **Activation map.** With the foundations decided, the channel, content, and team capability investment that now makes sense — a direction-setter for the quarter ahead, and the on-ramp to the work that follows. *Output: a one-page activation map.* - **Handed over as a working system.** The four layers hand over as structured context your team writes from and your AI tools read from, plus a brand linter that scores any piece of copy against the positioning. Brand consistency becomes something you can measure and delegate, instead of something one person has to police in Slack. *Output: a canonical context set and a working brand linter.* ### How it works Four phases, built to ship decisions — anchored by a strategic workshop where the key decisions get made with you. Typically three to five weeks, scoped to the moment. - **01 — Discovery & audit.** Audit of current brand, positioning, site, decks, and collateral; competitive scan; stakeholder interviews. Plus everything else you can throw at it — sales calls, support tickets, reviews, three years of how competitors have described themselves — read in full rather than skimmed. AI-moderated interviews run in parallel, so the picture reflects a wide set of voices rather than the few who could make the calendar work. *Output: a tight audit and the questions for the workshop.* - **02 — Strategic workshop.** The session where the big decisions get made — architecture and positioning, with the people who can actually decide. Before anything is committed, the calls are pressure-tested against a synthetic persona panel built from the real interviews, so the objections surface here rather than in a board meeting six weeks later. *Output: documented decisions with rationale.* - **03 — Synthesis & build.** Workshop and discovery turned into the full model — architecture, positioning, essence, messaging — plus the activation map. *Output: the foundation documents you run against for the next two quarters — as documents, and as the context set your tools read.* - **04 — Handover.** A walkthrough so your team can pick it up and apply it immediately, plus a follow-up window for questions as it goes live. ### The difference Strategy and systems, not a brand-agency retainer. Brand architecture and positioning sit at the core of the Inference practice, not as a side service — and the foundations are built to be run. - **The operating playbook.** CMO-level judgement on the calls that are hardest to reverse. Twenty years inside the function across SoundCloud, Pitch, Grover, and Prolific — positioning, brand, architecture, and messaging through every stage from seed to scale. - **Built as a system.** Foundations your team operates, not a slide deck they stop opening. Most brand strategy arrives as a deck, and decks lose — two quarters later nobody opens it and everyone describes the company slightly differently. This works the other way round. Everything you can throw at it gets read in full, AI-moderated interviews run alongside so the strategy reflects a wide set of voices, and the foundations hand over as working infrastructure that keeps your brand coherent. ### Who it's for - The business has moved — a new stage, a second product, a bigger team, fresh funding — and the brand strategy needs a refresh. - You need the foundational decisions made and documented, not advised on indefinitely. - You have a team (or you) who'll execute once the foundations are decided. - You'd rather decide the architecture deliberately now than reverse it later. It isn't for you if you want a logo or visual identity (this is strategy, not design), ongoing sparring (that's the Strategic Marketing Partner), someone to lead and run the function (that's Fractional CMO), or AI to generate your brand strategy — the machine widens the evidence, the judgement stays human. ### Frequently asked **How is this different from a brand agency?** An agency usually runs for months on a retainer and delivers identity and creative. This is one operator making the four foundational decisions with you, in weeks, and handing them back as documents you own. Strategy, not design — and no ongoing dependency. **What does "AI-native" actually mean here?** Three concrete things. The evidence goes wider: everything you can throw at it gets read in full, and AI-moderated interviews mean the strategy reflects a broad set of voices rather than the handful who could make the calendar work. The decisions get pressure-tested against a synthetic persona panel built from those interviews before they're committed. And the foundations hand over as a working system rather than a deck. What it doesn't mean: the machine writing your brand strategy. The judgement is the product, and it stays mine and yours — AI widens the evidence and changes what you can operate afterwards; it doesn't make the calls. **What if we already have some of this?** We start from what's there. The sprint audits what you've got, keeps what's working, and rebuilds only the layers that need it. **Do you do the logo and visual identity too?** No. This is brand strategy — architecture, positioning, essence, messaging. It gives a designer or agency a sharp brief to work from, but the sprint itself doesn't produce visual identity. **What happens after the sprint?** You own the documents and your team runs from them. If you want help bedding them in, the Strategic Marketing Partner is the natural next step; if the activation map surfaces a bigger piece of work, we can scope a follow-on. → Book a call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Library | https://thisisinference.com/library # Library The home for Inference's editorial and marketing-system families: Workflows, Outcomes and Guides. ### Library families - [Workflows](https://thisisinference.com/library/workflows): fixed-scope agentic automations for recurring marketing jobs, run in Ergo or installed bespoke. - [Outcomes](https://thisisinference.com/library/outcomes): fixed-scope marketing projects that transform the outcomes your marketing system is set up to deliver. - [Guides](https://thisisinference.com/library/guides): answer-first pages that prioritise what to automate. --- ## Workflows | https://thisisinference.com/library/workflows # Workflows Upgrade your marketing systems with agentic automations. The same self-learning flows that power Ergo, implemented anywhere. ### Workflow installations - [Competitor Intelligence Workflow](https://thisisinference.com/library/workflows/competitor-intelligence): an agentic workflow that watches named competitors, filters noise, and briefs your team on moves that touch a live decision. - [Weekly Marketing Report Workflow](https://thisisinference.com/library/workflows/weekly-marketing-report): an agentic workflow that collects named sources, compares them to target, and ships a weekly marketing pack with anomaly flags and a metric appendix. - [Marketing Board Update Workflow](https://thisisinference.com/library/workflows/marketing-board-update): an agentic workflow that assembles a decision-ready board-cycle update from live data, with a decision log and metrics appendix. - [Content Performance Review Workflow](https://thisisinference.com/library/workflows/content-performance-review): an agentic workflow that ranks published content from analytics, names winners and losers, and recommends what to make or refresh next. - [Lead Research and Scoring Workflow](https://thisisinference.com/library/workflows/lead-research-and-scoring): an agentic workflow that researches and scores new leads against your ICP before they reach the team, with routing and a queue digest. - [Sales-call Messaging Insights Workflow](https://thisisinference.com/library/workflows/sales-call-messaging-insights): an agentic workflow that extracts objections, customer language and messaging tweaks from sales calls. ### Install a workflow Three ways to get these workflows up and running in your system: sign up for Ergo and get the full agent experience, or, if you need something a little more custom, Inference can build bespoke installations for your environment. Or use these as inspiration to design your own, and let us know what you build! --- ## Competitor Intelligence Workflow | https://thisisinference.com/library/workflows/competitor-intelligence # Competitor intelligence An agentic workflow that watches named competitors, filters noise, and briefs your team on moves that touch a live decision. Watches your competitive set, filters the noise, and briefs your team when significant market changes are flagged. Runs in Ergo, as a bespoke install, or as a pattern to copy. Run automated market scans. Simply name your competitors and this workflow tracks their public surface (including news, website updates, customer stories, reviews), keeps a log of everything it finds, and briefs the team when a material change worth your attention happens. ### What it does - Positioning and messaging: homepage, campaigns, use cases. - Product and commercial changes: releases, packaging, pricing. - Company direction: news, leadership, partnerships, hiring. - Marketing activity: content, events, channels. ### Setup - Watchlist: the competitors that matter, and which public surfaces to track. - Cadence: how often the brief lands. Weekly is the default. - Decision questions: what the brief must answer: pricing, positioning, GTM, hiring bets. ### Ships - Signal brief: recurring digest of changes that need discussion. - Competitor move log: dated records with sources and assessment notes. - Quarterly pattern review: repeated activity and direction shifts over time. ### How it works 1. Watch: scan named public sources on cadence. Log each change with URL, date and source type. 2. Filter: deduplicate and drop one-off noise. Keep repeating patterns and material changes. 3. Assess: score remaining signals against the decision questions set at setup. 4. Brief: compile only material changes into the signal brief. ### FAQ - What is a workflow? A fixed-scope agentic automation for one recurring marketing job. Same loop, same outputs, configured to your market. - Ergo or standalone? In Ergo it runs as a live capability in Slack. Standalone is a bespoke install into your stack. Same workflow either way. - What do I bring? A competitor list, the surfaces worth watching, and the decisions the brief should support. Run in Ergo: https://thisisinference.com/ergo Book a scoping call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Weekly Marketing Report Workflow | https://thisisinference.com/library/workflows/weekly-marketing-report # Weekly marketing report An agentic workflow that collects named sources, compares them to target, and ships a weekly marketing pack with anomaly flags and a metric appendix. Deliverable-led weekly pack so humans spend time deciding, not wrangling data. Runs in Ergo, as a bespoke install, or as a pattern to copy. Run a weekly reporting loop. Name the sources, the KPIs and the questions the pack must answer. The workflow pulls the period, compares it to target and last week, flags anything past threshold, and ships a pack the meeting can use. ### What it does - Source pull: APIs, exports and sheets named at setup. - Period comparison: this week against last week and against target. - Anomaly flags: only the metrics that crossed your thresholds. - Pack assembly: report, charts, appendix, optional Slack summary. ### Setup - Sources: the APIs, exports and sheets the pack is allowed to read. - KPIs and targets: the numbers that count, and the line they have to beat. - Audience and cadence: who the pack is for, and when it lands. Weekly is the default. - Narrative questions: what the write-up must answer before anyone opens a spreadsheet. ### Ships - Weekly report: the pack leadership reads. - Anomaly flags: metrics that crossed the thresholds you set. - Metric appendix: the tables behind the narrative. - Optional Slack summary: the same story, short, in the channel. ### How it works 1. Collect: pull the named sources for the period: APIs, exports and sheets. Log coverage so missing sources show up. 2. Compare: compute period deltas against last week and against target. 3. Explain: draft the narrative against the questions set at setup. Flag anything past threshold. 4. Package: assemble the report, charts and metric appendix. Send the Slack summary if configured. ### FAQ - What is a workflow? A fixed-scope agentic automation for one recurring marketing job. Same loop, same outputs, configured to your market. - Ergo or standalone? In Ergo it runs as a live capability in Slack. Standalone is a bespoke install into your stack. Same workflow either way. - What do I bring? The sources, the KPIs and targets, who the pack is for, and the questions it has to answer. Run in Ergo: https://thisisinference.com/ergo Book a scoping call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Marketing Board Update Workflow | https://thisisinference.com/library/workflows/marketing-board-update # Marketing board update An agentic workflow that assembles a decision-ready board-cycle update from live data, with a decision log and metrics appendix. Decision-ready board-cycle update from live data. Runs in Ergo, as a bespoke install, or as a pattern to copy. Run a board-cycle loop. Name the cadence, the plan, the metric sources and the open asks. The workflow assembles the scorecard, writes the narrative, surfaces risks and asks, and packages an update the room can decide from. ### What it does - Scorecard: live metrics against the plan and OKRs. - Narrative: what moved, what held, what needs a decision this cycle. - Risks and asks: issues and requests that belong in the room. - Board pack: doc or deck, plus the appendix behind it. ### Setup - Cadence: when the board cycle lands. - Plan and OKRs: the commitments the update is scored against. - Metrics and finance sources: where the numbers come from. - Open asks: the decisions already in play. ### Ships - Board update: doc or deck the room reads. - Decision log: what was asked, what was decided, what is still open. - Metrics appendix: the tables behind the narrative. - Suggested asks: the requests the data supports. ### How it works 1. Assemble scorecard: pull live metrics and finance sources against the plan and OKRs. 2. Narrate: write what moved, what held, and why it matters for this cycle. 3. Surface risks/asks: flag risks and the decisions the board still owes, including suggested asks. 4. Package: compile the board update, decision log and metrics appendix. ### FAQ - What is a workflow? A fixed-scope agentic automation for one recurring marketing job. Same loop, same outputs, configured to your market. - Ergo or standalone? In Ergo it runs as a live capability in Slack. Standalone is a bespoke install into your stack. Same workflow either way. - What do I bring? The board cadence, the plan and OKRs, the metric sources, and the asks already in play. Run in Ergo: https://thisisinference.com/ergo Book a scoping call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Content Performance Review Workflow | https://thisisinference.com/library/workflows/content-performance-review # Content performance review An agentic workflow that ranks published content from analytics, names winners and losers, and recommends what to make or refresh next. What's working from data not vibes. Runs in Ergo, as a bespoke install, or as a pattern to copy. Run a content review loop. Point it at the inventory, the analytics and the success metrics. The workflow ranks what you published, names winners and losers, and recommends what to make, refresh or stop. ### What it does - Inventory ingest: what shipped in the period, from CMS and the URL list. - Attribution: traffic, conversion and assisted outcomes on each piece. - Scoring: ranked against the success metrics you named. - Recommendations: what to double down on, refresh, or cut. ### Setup - Content inventory and CMS: what shipped, and where it lives. - Analytics: the property that holds performance. - Success metrics: what working means for this programme. - Cadence: how often the review lands. ### Ships - Ranked table: every piece, scored against the metrics you named. - Winners and losers brief: what to keep doing, and what to stop. - Recommendations: next bets, with the evidence behind them. - Optional refresh briefs: rewrite briefs for pieces worth another pass. ### How it works 1. Ingest: pull the content inventory and the analytics for the period. 2. Attribute: attach performance to each URL or asset from the named property. 3. Score: rank against the success metrics set at setup. 4. Recommend: name winners, losers, and the next pieces to make or refresh. ### FAQ - What is a workflow? A fixed-scope agentic automation for one recurring marketing job. Same loop, same outputs, configured to your market. - Ergo or standalone? In Ergo it runs as a live capability in Slack. Standalone is a bespoke install into your stack. Same workflow either way. - What do I bring? The content inventory, the analytics property, the success metrics, and how often you want the review. Run in Ergo: https://thisisinference.com/ergo Book a scoping call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Lead Research and Scoring Workflow | https://thisisinference.com/library/workflows/lead-research-and-scoring # Lead research and scoring An agentic workflow that researches and scores new leads against your ICP before they reach the team, with routing and a queue digest. Scores new leads against ICP before they hit the team. Runs in Ergo, as a bespoke install, or as a pattern to copy. Run an intake loop on new leads. Give it the ICP rubric, the CRM, the enrichment sources and the thresholds. The workflow researches each lead, scores it, routes or briefs the team, and only escalates what is worth a human. ### What it does - Intake: new leads from the named CRM or source, as they land. - Research: firmographic and public-signal enrichment against the rubric. - Scoring: fit against ICP, with disqualifiers applied. - Routing: queue, CRM update, or escalation when the score warrants it. ### Setup - ICP rubric and disqualifiers: who counts, and who never should. - CRM and lead source: where new leads arrive. - Enrichment sources: the tools and public surfaces allowed for research. - Thresholds and routing: score bands, and where each band goes. ### Ships - Scored lead cards: fit, evidence, and the recommended next step. - CRM updates: score and notes written back where connected. - Queue digest: the period's scored leads, in one place. - Escalations: high-fit or odd cases pulled for a human. ### How it works 1. Intake: capture each new lead from the named CRM or source. 2. Research: enrich against the ICP rubric using the allowed sources. 3. Score: apply the rubric and disqualifiers. Record the evidence on the card. 4. Route/brief: write back to the CRM where connected, add to the digest, or escalate. ### FAQ - What is a workflow? A fixed-scope agentic automation for one recurring marketing job. Same loop, same outputs, configured to your market. - Ergo or standalone? In Ergo it runs as a live capability in Slack. Standalone is a bespoke install into your stack. Same workflow either way. - What do I bring? The ICP rubric and disqualifiers, the CRM or lead source, enrichment sources, and the score thresholds. Run in Ergo: https://thisisinference.com/ergo Book a scoping call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Sales-call Messaging Insights Workflow | https://thisisinference.com/library/workflows/sales-call-messaging-insights # Sales-call messaging insights An agentic workflow that extracts objections, customer language and messaging tweaks from sales calls. Extracts messaging insight from sales calls. Runs in Ergo, as a bespoke install, or as a pattern to copy. Run a call-review loop. Point it at the transcript source, the messaging framework and the segments you care about. The workflow codes themes, contrasts them with the current messaging, and briefs the team on objections, language and tweaks. ### What it does - Transcript ingest: calls from the named source, on cadence. - Theme coding: objections, proof points, and customer language. - Contrast: what buyers said versus the messaging framework. - Brief: the insight pack messaging and sales can share. ### Setup - Call and transcript source: where the recordings or transcripts live. - Messaging framework: the claims and language the calls are scored against. - Cadence: how often the insight brief lands. - Segments: which products, personas or stages to split. ### Ships - Insight brief: recurring digest of what buyers actually said. - Objection map: recurring pushbacks, with frequency and sample language. - Language bank: phrases customers use that messaging should steal. - Messaging tweaks: specific changes the calls support. ### How it works 1. Ingest: pull new calls or transcripts from the named source. 2. Code themes: tag objections, proof points and customer language. 3. Contrast vs messaging: score those themes against the framework set at setup. 4. Brief: compile the insight brief, objection map and language bank. ### FAQ - What is a workflow? A fixed-scope agentic automation for one recurring marketing job. Same loop, same outputs, configured to your market. - Ergo or standalone? In Ergo it runs as a live capability in Slack. Standalone is a bespoke install into your stack. Same workflow either way. - What do I bring? The call or transcript source, the messaging framework, the cadence, and any segments to split. Run in Ergo: https://thisisinference.com/ergo Book a scoping call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Outcomes | https://thisisinference.com/library/outcomes # Outcomes Fixed-scope marketing projects that transform the outcomes your marketing system is set up to deliver. ### Outcomes - [Run 2× more marketing experiments, no headcount increase](https://thisisinference.com/library/outcomes/run-2x-more-marketing-experiments): install an experiment design and shipping rhythm so volume rises without hiring. - Launch campaigns twice as fast (coming soon) - Double output without headcount (coming soon) - Cut marketing reporting time by 80% (coming soon) - Never start another campaign with a blank page (coming soon) - One strong idea becomes a month of marketing (coming soon) --- ## Run 2× more marketing experiments | https://thisisinference.com/library/outcomes/run-2x-more-marketing-experiments # Run 2× more marketing experiments, no headcount increase A fixed-scope project that installs an experiment design and shipping rhythm so your team runs twice the properly designed tests without adding people. Most marketing teams do not struggle to invent ideas. They struggle to turn them into properly designed experiments fast enough to learn. Tests pile up while the people who could ship them are already full, so volume stays stuck to headcount. This project installs the missing layer: a scored backlog, design kit, weekly cadence and learning log. Twice the properly designed experiments, without hiring, because the system carries the repetition and a person keeps the judgement. ### What you get - Experiment backlog and scoring - Design and measurement kit - Shipping cadence - Learning log ### How it works 1. Baseline current experiment volume and quality gaps. 2. Install the backlog, design kit and cadence. 3. Review until 2× volume is a habit. Book a call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Guides | https://thisisinference.com/library/guides # Guides Answer-first pages that help marketing leaders prioritise what to automate, then route into the right workflow or outcome. ### Guides - [Which marketing workflows should you automate first?](https://thisisinference.com/library/guides/which-marketing-workflows-to-automate-first) --- ## Which marketing workflows should you automate first? | https://thisisinference.com/library/guides/which-marketing-workflows-to-automate-first # Which marketing workflows should you automate first? By Thom Cummings. Updated 7 September 2026. Start with recurring jobs that have a clear definition of done, material business value, and bounded risk. ### Direct answer Automate first where the job repeats weekly, the output format is stable, and a person can still approve the decision. For most B2B marketing teams that means competitor monitoring, weekly reporting, experiment logging, and campaign briefing before open-ended content generation. ### Key facts - Prioritise repetition and clarity of done before novelty. - Score business value separately from how impressive the automation looks. - Match the job to assistant, automation, or adaptive agent. - Keep spend, positioning and product claims under human approval. ### Next routes - [Competitor intelligence workflow](https://thisisinference.com/library/workflows/competitor-intelligence) - [Run 2× more marketing experiments](https://thisisinference.com/library/outcomes/run-2x-more-marketing-experiments) - [Workflows hub](https://thisisinference.com/library/workflows) Book a call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Ergo — https://thisisinference.com/ergo # Smarter marketing systems, built together in your Slack. Ergo is an agentic marketing partner that maps strategy, pushes work forward, and helps teams make sharper calls. It uses Inference's frameworks, a team's accumulated context, and a human-in-the-loop for the judgement a prompt cannot provide. ### What Ergo can do - Run more than 30 specialist marketing capabilities, including positioning analysis, ICP mapping, channel plans, competitor scans, launch planning and AEO audits. - Start with an Opening Read to turn existing material into a shared strategic picture. - Keep decisions, hypotheses, learnings and priorities in memory from one conversation to the next. - Create decision documents, decks, briefs and repeatable playbooks. - Surface ideas through proactive routines, including Weekly Coach reviews and market-signal checks. - Connect through Composio to 1,000+ tools where marketing work and signal already live. - Escalate important questions to Thom Cummings or the right human, with the evidence and a clear recommendation prepared. ### How Ergo is tested Ergo is benchmarked against cold, general-purpose LLM baselines across eight criteria: call, diagnosis, grounded context, insight, evidence, challenge, boundaries/escalation and voice. The initial benchmark scored Ergo 77.75/100 against 60.75 for the strongest cold baseline. The test set is intentionally expanding. ### What's next (in build) - Ambient awareness — Ergo follows the conversation in a team's channels, not only messages addressed to it, and interjects proactively. - A client portal — one hub for the documents, decisions, accumulated context and in-flight work Ergo holds on an account. - Two new tiers — a lighter bring-your-own-key, operator-run tier without escalation, and a fully-managed agency tier with custom skills, sub-agents, builds and integrations. - Agent-to-agent mode — Ergo orchestrates a team's other agents, exposing what it knows about the business over MCP. - More execution — Ergo running campaigns, growth experiments and posting itself. - Recommended tools — a curated database of the martech stack Inference recommends, so Ergo can point at the right tool for the job. ### Private beta Beta places are a three-month commitment. Pricing is discussed on the first fit call. The beta includes frameworks, accumulated context, proactive routines and human escalation. → Apply: https://thisisinference.com/ergo#apply --- ## Agent teams — https://thisisinference.com/agent-teams # Marketing teams, rebuilt for the agent era. *A new collaboration format under collabs — agent team design.* A complete marketing team, designed and deployed in weeks — with agents in the right roles and humans where it counts. Built for founders who need a marketing team and marketing leads who need a bigger one. Scoped to your team. ### The shift The ground has shifted. Most marketing teams haven't caught up — not because the tools don't exist, but because nobody's designed the team around them. Agents aren't a layer you bolt on. They're a new kind of colleague, and they need the same thing any new hire needs: a role, a brief, a workflow, and someone who knows where they fit. The gap isn't tools. You've heard of Claude, OpenAI, Gemini, Perplexity. But a tool isn't a team — and most companies are using these individually, without any architecture holding them together. The gap is knowing which agent handles what, how they hand off work, where the human stays in the loop, and what "good" looks like when an agent produces it. That's org design. Every build draws on a tested library of agent configurations, skills, and workflow patterns — refined over hundreds of hours of real use. You're not paying for someone to figure this out from scratch. You're getting proven architectures, customised to your business. ### Who it's for **The founder — your marketing team, off the shelf.** For founders at seed to Series A who need a full marketing function but can't justify three hires. Full function from day one, a fraction of the cost of a single hire. **The head of marketing — a force multiplier for your existing team.** For marketing leads at scaling companies who need more output without more headcount. Hybrid org design — humans where they matter, multiplied output, same headcount. ### The process **01. Deep immersion** — strategic intake of business, goals, market, and how people actually work day-to-day. **02. Org design** — defining agent roles, skills, and handoffs, drawn from a library of proven configurations. **03. Build and deploy** — choosing and configuring the right platform (OpenAI, Claude, Gemini, or custom) and integrating with your existing tools (Slack, email, CMS, analytics). **04. Test and iterate** — running the team on real work, identifying failure modes, tuning agent behaviours. ### What you get - Fully configured agentic team — defined roles, briefs, and live workflows, operational from day one. - Org architecture documentation — clear map of who does what, how agents interact, where humans stay in the loop. - Integrations with your existing stack — Slack, email, CMS, analytics. - Handover session — full walkthrough so you can operate and expand the system yourself. - Human vs. agent clarity map. ### Packages A three-person marketing team costs £150–200k/year in salary alone — before tools, management overhead, and the months it takes to hire. We build yours in weeks. Pricing is scoped on a fit call. - **Early-stage.** A marketing team across your core channels — content, social, email and SEO; a complete build plan with every role, brief and workflow documented and handed to you; set up inside the tools you already use; a month of hands-on support after launch. - **Scaling.** Everything in Early-stage, plus: your full marketing function across every role the team needs, not just the core channels; humans and agents working as one team with clear handoffs; agents that brief and pass work to each other; the right platforms chosen and integrated across your stack. - **Enterprise.** Everything in Scaling, plus: a team built across the whole org, not just marketing; approvals, compliance and sign-off built into every workflow; architecture spanning multiple platforms and teams; stakeholder alignment and buy-in. ### Ongoing support An agentic team isn't set-and-forget. Models improve, tools evolve, business needs change. The retainer covers ongoing tuning and performance review, new agent roles as needs emerge, and a human eye on what agents produce. → Book a call: https://calendar.app.google/x4srBeZZxPu5jGkLA --- ## Agent Readiness Scorecard — https://thisisinference.com/quiz/agent-readiness # Where is your team on the AI maturity curve? *Quiz — agent readiness scorecard.* Find out where your marketing team really is on the AI maturity curve — and what to build next. 10 questions across five dimensions. A personalised report with your maturity score, dimension breakdown, and concrete agent opportunities. The scorecard is an interactive web app that runs in the browser. Taking it needs no account; after the 10 questions you enter your email to unlock the full personalised report, including a custom Agent Opportunity Map. For teams ready to build, the next step is the AI Readiness Workshop — a 2-hour remote session that maps your team's 3–5 highest-impact agents, sequences them for quick wins, and designs the oversight architecture, so you leave with a one-page build roadmap specific to your team. → Related: AI Readiness Workshop — https://thisisinference.com/collabs/ai-readiness --- ## Links - Substack (Annotations): https://annotationsbyinference.substack.com - LinkedIn (Thom Cummings): https://linkedin.com/in/thomcummings - Book a call: https://calendar.app.google/x4srBeZZxPu5jGkLA - Email: thom@thisisinference.com