In our last article, we introduced the idea that the people who benefit most from AI aren’t the most technical. They’re the most specified. We talked about specification debt at three levels: personal, professional, and organizational. And we briefly introduced a concept called Telos, a living document for specifying your life.
This is the promised deep dive.
Telos isn’t something we invented. The framework was created by Daniel Miessler, a security researcher and systems thinker who open-sourced it as a way to help individuals and organizations articulate their purpose with the kind of precision that both humans and AI can act on. His original framework focuses on creating “deep context” about things that matter. Problems, strategies, missions, projects, all connected and traceable.
What we’ve done is adapt and extend it for the specific challenge our clients face: how do I make my life, my work, and my goals legible to AI agents so they can actually help me?
This article walks you through building your own Telos file from scratch. By the end, you’ll have a living document that does two things: forces you to think clearly about what you want, and gives any AI tool you use the context it needs to stop giving generic advice.
What Telos is (and what it isn’t)
The word “telos” comes from Ancient Greek. It means purpose, end goal, ultimate aim. Aristotle used it to describe the final cause of something: the reason it exists.
A Telos file is a living document that captures who you are, what problems you’re trying to solve, and how you’re going about it. Every section flows from the one above. Your problems define your mission. Your mission shapes your goals. Your goals determine your projects. Nothing floats free.
What it is:
- A personal specification document, updated as your life changes
- A decision-making tool. When priorities compete, consult your Telos.
- A context document for AI agents. Load it into any AI session and the output quality shifts immediately.
- An honesty exercise. Writing it forces you to confront what you actually want versus what you say you want.
What it isn’t:
- A vision board. There are no affirmations here. Every section requires specificity.
- A productivity system. Telos doesn’t tell you how to manage your time. It tells you what your time should be spent on.
- A finished document. If your Telos file hasn’t changed in three months, you’re either not growing or you’re not updating it.
- A resume or personal brand statement. This is for you first. Everything else is secondary.
Miessler’s original insight was that over 99% of people aren’t “activated” because they’ve never articulated who they are or what they’re pursuing. That resonated with me. I’ve met plenty of smart, capable people who couldn’t tell you what they were optimizing for if you asked them directly. Not because they’re lost, but because they’ve never been forced to write it down.
The structure: top-down, each section earned
The power of Telos is in its hierarchy. You don’t start with goals. You start with problems. Everything derives from the level above it. If you can’t trace a project back to a goal, and that goal back to your mission, and that mission back to a problem you actually care about, that project probably shouldn’t exist.
Here’s the full structure. I’ll walk through each section with guidance on how to write it.
History
Where you come from. Not a resume. A context document.
This section exists because your history shapes your perspective, and your perspective shapes everything else. The decisions you make, the problems you notice, the approaches you favor. An AI agent reading your Telos should understand why you see the world the way you do.
Write 2-3 paragraphs covering:
- Your professional arc. Not every job. The trajectory and the turning points.
- The moment (if there was one) when your current direction became clear.
- Any context that shapes how you think: industry, geography, family situation, formative experiences.
Keep it honest. This isn’t LinkedIn. Nobody is grading you. The value is in being specific about how you got here.
Problems
What do you see in the world that bothers you enough to act?
This is the hardest section to write well. Most people either go too broad (“the world is broken”) or too specific (“my inbox is overwhelming”). The right level is somewhere in between: problems big enough to be meaningful, specific enough to be actionable.
Guidelines:
- Aim for 3-5 problems. More than that and you’re diluting focus.
- Each problem should be something you’ve personally observed, not something you read about and thought “yeah, that’s bad.”
- Use identifiers (P1, P2, P3) so you can reference them later.
- Frame them as gaps, tensions, or failures. “X is happening, and it’s causing Y.”
Example:
- P1 Most small businesses have no idea how AI could help them and no trusted source to ask.
- P2 The people who could benefit most from AI tools are the ones least likely to have the time or confidence to learn them.
- P3 My team spends 40% of their time on work that could be automated, but we’ve never mapped what that work actually is.
Notice these aren’t abstract. They’re observations from someone’s actual experience. That’s what makes them useful.
Mission
One sentence. What you’re trying to do about those problems.
This is where most people want to write a paragraph. Don’t. The discipline of a single sentence forces clarity. If you can’t say it in one sentence, you don’t understand it well enough yet.
Your mission should:
- Address at least two of your identified problems
- Be specific enough that you could explain it to a stranger in 10 seconds
- Be ambitious enough that it won’t be “done” in a year
Example: “Help small business operators understand and adopt AI practically, so they stop wasting time on work that machines should do.”
If you’re struggling, use this formula: Help [who] do [what] so that [outcome]. It’s not elegant, but it works.
Narratives
The beliefs that underpin your mission. Why this matters to you specifically.
Narratives are the stories you tell yourself about why your mission matters. They’re your operating assumptions. Writing them down is valuable because it surfaces beliefs you’ve never examined.
Aim for 3-5 narratives, each labeled (N1, N2, etc.):
Example:
- N1 The gap between AI-literate and AI-illiterate will define economic opportunity for the next decade.
- N2 Most people aren’t resistant to AI. They’re overwhelmed and afraid. What they need isn’t hype. It’s honest guidance.
- N3 The people closest to the work are the best judges of where AI can help. They just need frameworks, not vendor demos.
These aren’t facts. They’re beliefs. Some of them might be wrong. Writing them down lets you test them against reality over time.
Goals
Concrete objectives, tagged by domain.
This is where Telos gets practical. Goals are specific outcomes you’re pursuing, each tagged so you can see how your energy is distributed across the domains of your life.
Tagging convention:
[work]for your primary employment[side]for side projects, freelance, or business ventures[personal]for health, relationships, finances, growth
Guidelines:
- Each goal should be traceable to your mission. If it isn’t, ask why it’s here.
- Be specific. “Get better at AI” isn’t a goal. “Complete three client AI assessments in the next quarter” is.
- Include both ambitious goals and maintenance goals. Not everything needs to be a moonshot. “Be home for dinner four nights a week” is a legitimate goal.
- Use identifiers (G1, G2, G3) for cross-referencing.
Example goals:
- G1
[work]Automate the three most time-consuming monthly reports within 6 months. - G2
[work]Build an internal AI playbook that any team member can follow without my help. - G3
[side]Launch the consulting practice and sign 2 clients within 6 months. - G4
[personal]Exercise three times a week consistently for 90 days. - G5
[personal]Be home for dinner four nights a week, not two.
Look at the balance across tags. If everything is [work] and nothing is [personal], that’s information worth having.
Challenges
The real obstacles. Not the ones that sound good. The honest ones.
Most planning frameworks skip this section. Telos doesn’t, because challenges are the main reason goals don’t get met. If you don’t name them, you can’t address them.
Rules:
- Tag them the same way as goals.
- Be painfully honest. “I procrastinate on things that aren’t immediately rewarding” is more useful than “time management could be improved.”
- Name the pattern, not just the symptom. “I start projects with energy and abandon them when they get boring” is a pattern. “I didn’t finish the website” is a symptom.
Example:
- C1
[work]I don’t have enough AI knowledge to evaluate tools confidently. I default to what vendors tell me. - C2
[side]The consulting practice needs marketing, but marketing feels uncomfortable and I keep avoiding it. - C3
[personal]Financial pressure makes it hard to invest time in long-term goals. Every decision gets filtered through short-term survival.
The value of writing these down isn’t that it fixes them. It’s that it makes them visible. You can’t strategize against a challenge you haven’t named.
Strategies
How you’re addressing each challenge. Concrete approaches, not aspirations.
Every strategy should map to at least one challenge. If you have a challenge with no strategy, that’s fine. It means you haven’t figured it out yet. That’s honest. But don’t write a vague strategy just to fill the space.
Example:
- S1
[work]Dedicate Tuesday mornings to AI learning. Start with our actual workflows, not abstract concepts. Block the calendar. No meetings. - S2
[side]Treat marketing as a system: batch content creation on Sundays, schedule for the week. Remove the daily decision of “should I post something?” - S3
[personal]Track expenses weekly instead of monthly. Smaller intervals make the numbers less overwhelming and decisions more immediate.
Notice: these are behaviors, not wishes. “Be more disciplined” isn’t a strategy. “Block Tuesday mornings and protect the time” is.
Projects
The active work you’re doing right now. Each traceable to a goal.
Projects are the most concrete layer. They’re the actual initiatives you’re spending time on this month.
Example:
- Pr1
[work]Q2 Report Automation - Map the three most time-consuming reports, identify automation candidates, build prototypes. (Supports G1) - Pr2
[side]Consulting Website Launch - Portfolio, service descriptions, booking flow. Ship by end of month. (Supports G3) - Pr3
[personal]Morning Routine Rebuild - Wake at 6, exercise, review Telos before starting work. 30-day trial. (Supports G4)
When a project is done, mark it as complete. When it’s abandoned, note why. Both are valuable data.
Metrics
How you measure progress. Numbers, not feelings.
For each major goal, define at least one measurable indicator:
- M1
[work]Hours per month on manual reporting. Current: 64. Target: under 15. - M2
[side]Consulting clients signed. Current: 0. Target: 2 within 6 months. - M3
[personal]Evenings home for dinner per week. Current: 2. Target: 4.
You don’t need to track these obsessively. Update them monthly. The point is that when you review your Telos, you can see whether things are actually moving.
Wisdom
Lessons you’ve learned about yourself. Patterns you’ve noticed.
This is the section most people don’t expect to find in a framework like this. It’s also the section that becomes the most valuable over time.
Write down the things you know about yourself that took years to learn:
- “I do my best thinking in the morning. Meetings before 10am cost me my best hours.”
- “Starting projects is easy. Finishing them is not. The pattern is clear: momentum dies when the work gets uncomfortable.”
- “Consistency beats intensity. Showing up regularly with moderate effort compounds. Bursts of motivation don’t.”
These entries are constraints your AI agents should know about too. If an agent knows you abandon projects when they hit the marketing phase, it can be designed to break marketing tasks into smaller, less aversive chunks.
Journal
A running log. What happened, what you learned, what shifted.
Short entries. Date-stamped. Not daily. Just whenever something significant happens: a project ships, a strategy fails, a priority changes, a lesson crystallizes.
The journal creates a timeline of your thinking. When you review it quarterly, you’ll see patterns that aren’t visible day-to-day.
How Telos changes your relationship with AI
Here’s where the specification engineering angle becomes concrete.
Before Telos: You open an AI chat. You ask for help with something. The agent gives you generic advice based on whatever you typed in the prompt. You spend time correcting, redirecting, providing context that the agent should have had from the start.
After Telos: You load your Telos file at the start of a session. The agent immediately understands your mission, your constraints, your current projects, and what “good” looks like for you. It stops suggesting things that don’t fit your life. It starts suggesting things that do.
This isn’t theoretical. When I started loading my own Telos into AI sessions, the difference was immediate. The agent stopped recommending productivity frameworks I’d already tried. It stopped suggesting strategies that conflicted with my constraints. It started operating with the context of someone who actually knew me.
Practical applications:
Decision support. “I’m considering taking on a new project. Based on my current goals and challenges, does this align or distract?” An agent with your Telos can give a grounded answer instead of a generic one.
Weekly planning. “Based on my current projects and strategies, what should I focus on this week?” The agent can cross-reference your goals, challenges, and active projects to suggest priorities that actually match.
Strategy evaluation. “I’ve been working on S2 for two months and it’s not producing results. What adjustments would you suggest given my constraints?” The agent can reason about your specific situation instead of offering textbook advice.
Accountability. “Review my metrics and tell me where I’m falling behind. Be direct.” An agent with your baseline numbers can give you an honest progress report.
The key insight is that your Telos file isn’t just self-reflection. It’s infrastructure. It’s the context layer that transforms a generic AI assistant into one that understands your specific life.
Common mistakes and how to avoid them
After building my own Telos and helping others start theirs, I’ve noticed consistent patterns in where people struggle:
Writing aspirations instead of specifications. “I want to be a better leader” belongs in a journal entry. “Increase direct report satisfaction scores from 3.2 to 4.0 by Q4” belongs in a Telos file. Every entry should be specific enough that someone else could evaluate whether you’ve achieved it.
Making it too long. Your first draft will probably be 3-4 pages. That’s fine. Then edit ruthlessly. The goal is a document concise enough to load into an AI context window without crowding out the task at hand. Aim for 2-3 pages once it’s mature.
Never updating it. A Telos file that hasn’t changed in three months is either perfect (unlikely) or abandoned. Set a monthly reminder to review it. Life changes. Goals shift. Projects finish. New challenges emerge. The document should reflect where you are, not where you were.
Skipping the challenges section. This is the most common omission and the most costly. Without named challenges, your strategies have no targets and your goals have no honest assessment of what stands between you and achieving them.
Making it performative. If you’re writing your Telos to impress someone, you’re doing it wrong. The most useful entries are the uncomfortable ones. The challenge you’re embarrassed about. The goal you’re afraid you’ll fail at. The pattern you keep repeating. Those are exactly the entries that make the document valuable.
Getting started: your first Telos in 30 minutes
You don’t need to write the whole thing today. Start with the core:
1. Problems (5 minutes). Write three problems you’ve personally observed that bother you enough to act. Not world problems. Your problems. The gaps you see in your work, your industry, your daily life.
2. Mission (5 minutes). One sentence. What are you trying to do about those problems? Use the formula if you’re stuck: Help [who] do [what] so that [outcome].
3. Goals (10 minutes). Write five goals, tagged [work], [side], or [personal]. Be specific. Include at least one personal goal.
4. Challenges (5 minutes). Write three honest challenges. The real ones. The ones you’d normally keep to yourself.
5. One project (5 minutes). Pick the most important thing you’re working on right now. Write it down with a one-line description and connect it to a goal.
That’s your starter Telos. Save it as a text file somewhere your AI tools can access it. Load it at the start of your next AI session and notice what changes.
Then come back to it next week. Add the History section. Add Narratives. Add Strategies for your challenges. Add Metrics for your goals. Let it grow organically.
The document will never be finished. That’s the point. It’s a living specification of a living person.
The uncomfortable gift
I’ll be honest: writing my first Telos file was uncomfortable. The problems section came easily. The mission felt right. But the challenges section forced me to write down things I’d been avoiding. Patterns I’d rather not name. Constraints I’d been pretending didn’t exist.
That discomfort is the point.
Specification, whether it’s for software, for an organization, or for your life, requires confronting ambiguity and replacing it with clarity. That process is never comfortable. But it’s always valuable.
Daniel Miessler said something when he published his own Telos that stuck with me: “This feels very strange to put on the internet, because of how intimate it is. But it’s worth it if it gets other people to do something similar.”
Your Telos doesn’t need to go on the internet. It doesn’t need to be shared with anyone. But it does need to be written. Because the act of specifying who you are and what you want turns out to be the foundation for everything else: better decisions, better AI output, and a clearer sense of where you’re going.
The specification is the work. And this one starts with you.