Who is Trail for?

Trail is a general-purpose AI memory and second-brain platform. It is useful any time someone has a body of knowledge that (a) grows over time, (b) deserves to be cross-referenced rather than thrown into a chat window once per question, and (c) should answer back accurately, in your own voice, with citations.

The profiles below are the ones we see most often. Each one ends with the concrete payoff — what changes the day you stop "asking ChatGPT generic questions" and start "asking your own Trail".

Software engineering teams using AI agents

The setup. You are running Claude Code, Cursor, Codex, or another AI coding agent against a real codebase. Every session burns through the context window before the agent finishes the task. The decisions you made yesterday — "we picked Postgres over MongoDB because of write-heavy load", "this bug only repros on iOS Safari with service-worker caching" — vanish the moment the chat compacts.

What you put into Trail. ADRs (architecture decision records), caught bugs with their root-cause analyses, conventions established mid-session ("all new LLM calls go through spawnClaude, never fetch"), interop quirks between services, performance findings, prompt-engineering wins, configuration gotchas.

What Trail compiles. Each session writes Neurons via Trail's MCP server (mcp__trail__write for Claude Code; equivalent tools for Cursor + Codex). The Neurons are stable, citable, deduplicated by canonical seqID (buddy_00000049, trail_00000163), and cross-linked: the Neuron about your Postgres choice cites the Neuron about write-load benchmarks and is cited back from the Neuron about schema migrations.

What you get back. The next session — minutes or weeks later — starts by reading relevant Neurons via mcp__trail__search or mcp__trail__recent_activity. Caught bugs surface before they are re-caught. ADRs surface before someone unknowingly contradicts one. "We've solved this before, but where?" stops being a guessing game.

Concrete payoff. No more re-deriving decisions every Monday. Multiple agents (Claude Code + Cursor + a planning Claude Desktop) share one memory; what one agent learns, the others can read. The context window stops being the limit on continuity.

Reflexologists and Traditional Chinese Medicine practitioners

The setup. You have years of training, your own course notes, treatment protocols by indication, anatomy charts, meridian and point references, books in two or three languages, and notes from cases that worked unusually well. None of it is searchable in any meaningful way. When a client asks "what about this combination?", you go from memory.

What you put into Trail. Your textbooks (PDF — Trail extracts text + image content via Vision), your typed treatment protocols, scanned handwritten notes (OCR'd), client-specific learnings (kept as private user-notes on the relevant Neurons, not visible in chat unless you opt in), course slides, audio recordings of seminars (transcribed automatically), and reference glossaries of points, meridians, treatments, and conditions.

What Trail compiles. A graph of Neurons where treatment protocols cite specific points, points are part of meridians, conditions reference contraindications, and supersedence edges capture "this 2024 guidance replaces the 2018 protocol". A custom schema declares your domain vocabulary so Trail's ingest pipeline knows that "GB-21" is a point, that "Lung meridian" is a meridian, and that "frozen shoulder" is a condition.

What you get back. Chat: "What treatment combination should I consider for a client with chronic frozen shoulder who also has a known sensitivity to deep pressure?" → grounded answer pulling from your own protocols, with citations to the exact Neurons. A patient education widget on your website (optional) lets new clients ask basic questions and get answers in your voice, citing your own material.

Concrete payoff. Twenty years of accumulated practice becomes one searchable, askable surface. Your "Din tanke" reflections — the private notes that would never make it into a published textbook — become part of your own consultation tool, opt-in shareable when they should be.

Physiotherapists and movement specialists

The setup. Exercise libraries, anatomical references, peer- reviewed research papers, evidence-based protocols by condition, patient education sheets, course materials from CPD, your own clinical observations. You re-explain the same exercise rationale to fifty patients a year and you keep finding yourself searching for that specific paper that justified the timing of the eccentric phase.

What you put into Trail. Research papers (PDFs — Trail extracts methodology, results, and discussion sections into separate Neurons that cite each other), exercise protocols with anatomical reasoning, patient handouts, your CPD certifications and notes from courses, treatment outcome measurements, your own published or unpublished case studies.

What Trail compiles. Neurons for individual exercises (with indication, contraindication, dosage, progression), Neurons for conditions (with their typical evidence-based interventions), Neurons for principles (eccentric loading, motor control, pain neuroscience education), and bidirectional edges that let you traverse from "patient with ACL graft at week 6" to "exercises appropriate for that phase" to "papers supporting the timing".

What you get back. Patient education becomes one ask: "Generate a one-page handout for a patient at week 6 post-ACL with focus on quadriceps activation, citing our standard protocol." Continuing education stops feeling like running on a treadmill — every new course's notes get integrated into the same graph, with contradictions flagged automatically (Trail's lint pass catches "this 2025 paper contradicts our 2022 protocol — should we update?").

Concrete payoff. The years of professional learning that typically live in shelves and Dropbox folders become an active consultation surface. New evidence integrates without overwriting your accumulated clinical wisdom — supersession is explicit, not silent.

Wellness coaches

The setup. You blend several disciplines — nutrition, behavioral psychology, sleep science, stress physiology, breathwork, maybe somatic practices. Your sources cross fields: Huberman podcasts, peer-reviewed reviews, course materials from your certifications, books you have annotated, client-session notes, coaching frameworks. You translate this into client-friendly language a hundred times a year.

What you put into Trail. Podcast transcripts (audio in → transcribed → ingested), book annotations, certification course materials, coaching frameworks (NLP, Motivational Interviewing, SFBT, etc.), nutrition references, sleep-science summaries, your own client-anonymized case notes, the prompts and frameworks you have refined for specific challenges (sleep, weight, anxiety, habits).

What Trail compiles. A Neuron graph that cross-references nutrition advice to behavior-change frameworks to sleep science to motivational technique. When you mention "low-glycemic load" in one Neuron and "blood-sugar variability" in another, the wiki-link resolves automatically and the graph traversal lets you ask multi-discipline questions ("What's the connection between post-meal walks and decision fatigue?") and get a synthesised answer with citations across your sources.

What you get back. Client-prep becomes "Ask Trail: what do I already know about a client situation like this one?" — not "let me search my Notes app for fragments". Generating actionable client homework draws on your full library, in your voice, every time. Group programs and online courses can be authored faster because the source material is already organized as a queryable second brain.

Concrete payoff. Your eclectic, multi-disciplinary practice stops feeling like a stack of disconnected certifications and becomes one coherent worldview that compiles every time you add to it.

Health professionals (nurses, GPs, allied health)

The setup. Clinical guidelines change. Drug interactions are remembered or not. Continuing education is mandatory and constantly adds new sources you may or may not look at again. Patient education materials need to be authoritative, current, and in plain language.

What you put into Trail. National clinical guidelines (NICE, local equivalents), CPD course materials, your specialty's seminal papers and reviews, drug-interaction references, patient-education templates, your own specialty-specific protocols, peer-reviewed journal subscriptions you actually skim.

What Trail compiles. A graph that knows which guideline supersedes which earlier version (supersession edges fire automatically when newer Neurons reference older ones), where specific drug interactions are documented, and which patient- education materials match which conditions. The lint pass flags contradictions between sources (a critical signal — "the latest hypertension guideline contradicts what we taught in last year's CPD").

What you get back. Point-of-care queries: "Is there a contraindication between [drug A] and [drug B] documented in any guideline I have ingested?" Patient handouts in plain language: "Generate a handout explaining post-operative wound care for a patient at a 5th-grade reading level, citing the relevant local guideline." CPD becomes cumulative rather than something that slips through your fingers between annual renewals.

Concrete payoff. Less time hunting through three different PDFs to confirm a guideline detail. Better-grounded patient education. A trail of which guidelines you have actually internalised — useful when you sit a competence renewal and want to know what your evidence base looks like.

Coaches more broadly (executive, business, life)

The setup. Your value is your library of frameworks + your ability to apply them to specific clients. The frameworks come from books, courses, mentors, and your own experimentation. Most of the leverage is in the connections between them — which framework fits which client situation.

What you put into Trail. Books you have studied (digital or audiobook → transcribed), framework decks, certification course notes, your own modified versions of frameworks, coaching session notes (anonymized), client-archetype profiles, your prep notes from cases that felt unusually clear or unusually stuck.

What Trail compiles. Frameworks as Neurons, archetypes as Neurons, signals as Neurons. Edges relate "client shows X signal" → "consider framework Y" → "watch for failure mode Z". Your own experimental adaptations become first-class Neurons that cite the original framework via "is-a" or "supersedes" edges, preserving lineage.

What you get back. Pre-session prep: "Given what I know about this client (private user-notes on their archetype Neuron), what framework am I underusing? Cite to my own materials." Group program design draws on the full library. Course or book authoring is a dialogue with your own compiled knowledge, not a search across folders.

Concrete payoff. Coaching becomes a deliberate practice with a memory. Your accumulated insight stops disappearing into session-notes folders and starts compounding.

Researchers, academics, and writers

The setup. Papers, drafts, lab notebooks, citations, half- finished essays, slide decks, email threads with co-authors, conference notes. Most of it lives in folders organised by date or project, not by idea.

What you put into Trail. Papers (PDF → text + figure extraction), your draft chapters, your own published work, seminar notes, citation database, conference programmes, the dozens of "this is interesting"-tagged things you saved over the year and never came back to.

What Trail compiles. Concept-Neurons that cite specific papers, paper-Neurons that link to the concepts they argue, contradiction edges where two papers disagree, supersedence edges where a newer paper updates an older one. The graph is the research-network you wished you had been keeping all along — except now it builds itself.

What you get back. Citation finding for a paper you are writing: "What have I read that supports the claim that X causes Y?" Reading queue management: "Which of the 47 papers tagged 'maybe relevant' are actually cited by my drafts now?" Cross-disciplinary connections: "Show me Neurons that connect [domain A] and [domain B] — what's the bridge?"

Concrete payoff. Writing becomes synthesis from a memory that has actually retained what you read, rather than a panicked search through your hard drive.

Consultants and boutique agencies

The setup. You have a methodology. You have client deliverables. You have research that informs your point of view. You have proposals you have written. Each engagement is bespoke but draws on the same intellectual stock.

What you put into Trail. Methodology documentation, sanitised case studies, proposals (templates and shipped), industry research you reference, your own thought-leadership writing, slide decks, the post-mortems where you actually learned something the methodology didn't capture.

What Trail compiles. Methodology Neurons cite case-study Neurons that demonstrate them in practice. Case studies cite the industry research that justified the recommendation. Post-mortems supersede or contradict earlier methodology Neurons, capturing learning explicitly. Proposals compose from methodology + relevant case studies in a shape clients can read.

What you get back. Proposal drafting goes from "search the shared drive" to "ask Trail to assemble the relevant case studies and methodology citations for a [vertical] [problem-type] engagement". New consultants onboard against a structured second- brain instead of a folder hierarchy. Brand voice stays consistent because the chat output draws from your own writing.

Concrete payoff. Tribal knowledge becomes institutional without ceasing to feel personal. The consultant who leaves takes a piece of expertise; the consultant who joins reads the Neurons.

Authors, thinkers, and the Luhmann audience

The setup. You have a notebook (or twelve). You have read a lot. You think for a living. Your ideas live in fragments across years of journals, drafts, marginalia, screenshots of tweets, voice memos.

What you put into Trail. Books with annotations, your own journals (typed or scanned), drafts of essays, voice memos transcribed, screenshots ingested via Vision, the half-finished manuscripts you keep meaning to come back to.

What Trail compiles. A Zettelkasten built by an LLM that respects your own framing. Each idea becomes a focused Neuron that cites the source it came from and links to adjacent ideas. The "Din tanke" (your own reflection) layer is private by default — your own thinking is preserved alongside the source material it reacts to, but doesn't leak into chat unless you opt in per Neuron.

What you get back. Writing a new piece becomes a conversation with your own past thinking. A search-friendly Zettelkasten without the discipline cost: you don't have to hand-write every Neuron — Trail does the compile, you do the curation.

Concrete payoff. The book-or-essay that has been "in your head for years" stops being abstract. Your accumulated reading becomes a tool that can answer back, not a wall of unread spines.

Small and medium businesses

The setup. Slack threads with the answer. Notion pages no one remembers. The team wiki that drifted out of date in 2023. The "who knows about X?" question that gets pinged to whoever answered it last time.

What you put into Trail. Existing Notion / Confluence / Slack-export content, vendor contracts, internal SOPs, post- mortems, customer-support knowledge, onboarding docs, decision logs.

What Trail compiles. A wiki that stays consistent because the lint pass surfaces orphans, contradictions, and stale content. New onboardings ingest into the same KB; the "institutional memory" actually accumulates instead of ossifying.

What you get back. New hires find answers without pinging seniors. Senior staff get pinged less. The "where is that contract?" search converges on a single source. Customer support gets an internal Trail to consult; sales gets one too.

Concrete payoff. Less context-switching for senior staff. Faster onboarding. Decisions stop having to be re-explained.

Customer-facing AI on your website

The setup. You have a brand, a voice, and accumulated content. Generic chat-bots ground their answers in the open internet — not your offering, not your tone, frequently wrong.

What you put into Trail. Your product documentation, your positioning, FAQs, blog posts, sales materials, customer-support templates.

What Trail compiles. A KB that becomes the grounding source for a brand-aware AI assistant. The chat widget on your site hits Trail's POST /api/v1/chat with a tenant-scoped bearer token; answers cite the specific Neuron that grounded them; the voice is yours because the source material is yours.

What you get back. A chat widget that answers your actual questions, not generic ones. Sales-qualified leads from the chat are passed to a human with the conversation context. Conversion-killing hallucinations stop happening — if Trail does not have the answer, it says so.

Concrete payoff. Fewer support tickets that ask things your docs already answer. Higher conversion on landing pages because the chat closes the "is this for me?" gap. Brand-safe AI on the front page — finally.

How to know if Trail is right for you

A pragmatic checklist. Trail earns its keep when most of the following are true:

• The same fact appears in multiple sources in your library, and you would prefer one canonical answer.
• Your knowledge evolves — last year's truth is sometimes this year's mistake, and you need supersession to be explicit.
• Provenance matters. "Where did I learn that?" is a question worth answering, whether for compliance, citation, or just intellectual honesty.
• Some kind of curator-in-the-loop is welcome (or required) — you don't trust an LLM to silently rewrite your knowledge base.
• You expect the system to get better the longer you use it, not flatter.

If most of those describe you, the next step is the quick start — five steps to a working Trail integration — or the deeper intro on what Trail is and isn't compared to RAG, NotebookLM, and similar tools.