Writing from the engine.
-
The 90-minute website audit you can run yourself
This is the audit we run on discovery calls, converted into a timed solo exercise: 15 minutes on crawler access, 25 on extractability, 20 on identity signals, 20 asking the engines directly, and 10 building a punch list ordered by effort. Most fixes need an editor, not an engineer.
Jul 18, 2026 -
Case note: the F-grade site that hit a B in fourteen days
An HVAC contractor ran our free homepage grader and scored 31 out of 100, with sixteen of 22 checks failing. Fourteen days of fixes later the site regraded at 82, and Perplexity began citing it for local service queries. Here is the day-by-day log.
Jul 17, 2026 -
Entity SEO: making your brand a thing machines know
If ChatGPT cannot say what your brand is, you have an entity problem, not a content problem. Here is the process we run: a five-engine baseline test, one canonical sentence, Organization schema with sameAs links, a parseable about page, and the 60-day timeline it honestly takes.
Jul 16, 2026 -
Prompt injection is a supply-chain problem
Prompt injection is not a prompt-writing problem. Every page, email, and PDF your agent reads is an unvetted dependency with a stranger as maintainer. Here is the supply-chain posture we run: mapped inputs, least-privilege tools, a quarantine airlock, and human gates on the irreversible three.
Jul 15, 2026 -
Engine log: the week the palette learned restraint
Bounce rose six points where our engine's hero animations got loud, so the system cut its own motion and palette theatrics across nine sites — and wrote the restraint into policy. A dispatch on calm as a learned behavior, with the numbers that taught it.
Jul 14, 2026 -
The client report nobody reads — and the one-line memo they do
We spent six hours a month on a fourteen-page client report that earned 47 seconds of attention. We replaced it with a one-sentence memo — change, cause, next move — and reply rates hit 78%. Here is the formula, three real examples, and the two times a year the long report still matters.
Jul 13, 2026 -
How answer engines pick between two similar sources
When two pages make the same claim, an answer engine cites one and ignores the other. We ran 60 paired buyer prompts across three engines to find the tie-breakers. Four signals decided almost every head-to-head, and most are fixable in an afternoon.
Jul 12, 2026 -
Case note: 30 days of per-client learning on a Shopify store
A nine-person Shopify home fragrance brand turned on per-client learning June 1. Thirty days later: organic sessions up 18 percent, AI citations up from 2 to 19 a month, edit rate down from 70 to 12 percent, and the owner's content time cut from six hours a week to 50 minutes.
Jul 11, 2026 -
The compounding content queue: never miss a publishing day again
Publishing daily is an inventory problem, not a discipline problem. A 14-day queue of finished posts absorbs sick days, launches, and client crunches, refilled by two 90-minute batch sessions a week. Here is the full system, including the 20-minute weekly review that runs it.
Jul 10, 2026 -
FAQ blocks that get quoted: structure, length, phrasing
Pages with a properly built FAQ block get cited by AI engines about 2.4 times as often as matched pages without one. The specification that earns those citations: questions phrased the way people ask an AI, answers of 40 to 90 words, verdict-first openings, and schema that matches the visible text.
Jul 9, 2026 -
Tool use is the new API design
Agent reliability is capped by tool quality, not model quality. We swapped models three times without moving task success, then redesigned our tools twice and completion jumped from 71 to 92 percent. Six design rules for naming, descriptions, errors, scope, retries, and instrumentation.
Jul 8, 2026 -
Engine log: teaching the brain to write in a client's brand voice
This week Avakata's engine onboarded a new brand voice: 40 samples ingested, a 14-feature fingerprint, twelve drafts, ten human grades, one rollback, and a voice-match score that climbed from 61 to 88. Here is the dispatch, including the two regressions.
Jul 7, 2026 -
When not to automate: a decision framework
Most automation advice points one direction: automate more. The expensive mistakes run the other way. This is the four-test framework we use before automating anything — frequency, variance, cost of error, and process stability — plus the break-even arithmetic that kills half of all automation ideas on contact.
Jul 6, 2026 -
The solopreneur's playbook for Claude Fable 5: what $10/$50 pricing actually buys you
Claude Fable 5 is globally available again as of July 1, 2026, priced at $10 per million input tokens and $50 per million output. Here is the concrete math for a one-person content and marketing operation, where the model earns its place, and when cheaper models still win.
Jul 5, 2026 -
Claude Fable 5 is back worldwide: the two-week ban, what it is, and why it matters
Anthropic launched Claude Fable 5 on June 9, 2026, US export controls pulled it from much of the world on June 12, and on July 1 it went back to global availability. Here is the timeline, the $10/$50 pricing, the restricted Mythos 5 twin, and what a Mythos-class model changes for a one-person operation.
Jul 5, 2026 -
Being the source: writing statistics AI engines quote
AI engines lift sentences that carry specific figures far more often than sentences that carry opinions. Here is how to write statistics engines quote: one liftable sentence, a stated denominator, a visible date, and a quarterly habit of publishing numbers only you can measure.
Jul 5, 2026 -
Case note: the booking conversation that outconverted the contact form
A solo brand strategist replaced her nine-field contact form with a scoped booking conversation. Over six weeks, booked calls went from nine to 31 on flat traffic, no-shows fell from a third to 13%, and four prospects became clients. Here is the before and after, and what the first version got wrong.
Jul 4, 2026 -
Agent memory: what to persist, what to forget, what to never store
More memory makes agents worse, not smarter. Our seven production agents share about 11,000 words of persistent memory, sorted by three lists: persist decisions, rules, and outcomes, forget transcripts and drafts on a timer, and never store secrets, client confidences, or one-off exceptions.
Jul 3, 2026 -
Freshness signals: why updatedAt is the cheapest citation lever
AI engines read your page's updated date in four places: the visible stamp, dateModified schema, sitemap lastmod, and the HTTP header. When all four agree and the update is real, citations follow. Here is the plumbing, the honesty rule, and the monthly pass that costs 20 minutes a page.
Jul 2, 2026 -
Q3 capacity math for a solopreneur running agents
A quarter is not 13 weeks. It is about 330 focused hours after vacation, admin, and reality. Here is the capacity math we run every July 1: price the agent maintenance tax first, allocate hours by fixed ratio, cap client count with a formula, and recount in week seven.
Jul 1, 2026 -
Engine log: June in review — what shipped, what learned, what got reverted
The June dispatch from the engine that runs avakata.agency: 58 changes shipped, 41 proposals held, nine reverts, and 13 hours of human time for the whole month. What the experiments taught us, where cycles got wasted, and what sits in July's queue.
Jun 30, 2026 -
How AI crawlers actually read your site (and what they skip)
We read 90 days of server logs across seven sites to see how GPTBot, ClaudeBot, and PerplexityBot really behave. They fetch raw HTML, mostly skip JavaScript, read shallow, and discard your nav, hero, and footer. Here is what they keep, and the five-minute audit that shows what they see.
Jun 29, 2026 -
Human-in-the-loop without becoming the bottleneck
Most human-in-the-loop setups review everything and scrutinize nothing. We run a dozen agents on about 35 minutes of review a day by approving classes of actions instead of individual ones. Here is the three-tier system, the batching schedule, and the monthly audit that keeps it honest.
Jun 28, 2026 -
The one-person agency org chart, drawn as software
Avakata's org chart has nine boxes. Two contain humans. The other seven are software — named roles with job descriptions, deliverables, and weekly reviews. Here is the chart box by box, what each role costs, and why drawing it before picking tools is the decision that matters.
Jun 27, 2026 -
Case note: reviving a dead service page with answer-first rewrites
A regional accounting firm's core service page had produced nine visits and zero leads in ninety days. Eight weeks after an answer-first rewrite, it was cited by Perplexity on four of twelve tracked queries, organic sessions were up 61%, and it had produced three qualified inquiries.
Jun 26, 2026 -
The five schema.org types that actually move AI citation
Schema.org lists more than 800 types. Five of them do measurable work for AI citation: Article, Person, Organization, FAQPage, and BreadcrumbList. Twelve months of citation audits say the rest is mostly decoration. Here is what each of the five carries and the fields that matter.
Jun 25, 2026 -
Engine log: what the bandit learned from 10,000 impressions
Our engine has run a Thompson-sampling bandit on the homepage hero for 41 days and crossed 10,000 impressions on Tuesday. Six variants went in, two survive, and the winner is not the one we would have picked. This is the log of what it tried and learned.
Jun 24, 2026 -
Pricing agentic work when the labor is robotic
When an agent does the work, the marginal hour costs about forty cents of tokens, and hourly billing collapses. Here is the pricing model we landed on after three wrong ones: flat monthly fees set between a cost floor and a replacement ceiling, with the math shown.
Jun 23, 2026 -
Six agent failure modes and the guardrails that catch them
Every production agent we have run has failed in one of six repeatable ways. Here are all six — confident wrong answers, runaway loops, silent drift, scope creep, poisoned handoffs, stale context — with the specific guardrail that catches each and what it costs to build.
Jun 22, 2026 -
The llms.txt file: what it is and exactly what to put in it
llms.txt is a curated markdown map of your site, served at your domain root for AI systems to read in one request. Ours is 41 lines and regenerates on every deploy. Here is what the file is, what it is not, and exactly what belongs in yours, line by line.
Jun 21, 2026 -
The AI solopreneur's weekly operating rhythm
A solopreneur with an AI stack still needs a weekly rhythm. Without one, the stack runs but the business drifts. Here is the exact weekly operating rhythm we run at Avakata — eight hours of human time, the rest on agents.
Jun 20, 2026 -
How to use AI to build a personal brand that gets cited
A personal brand in 2026 is not a social media presence. It is a citation profile: a body of attributed, practitioner content that AI engines recognize as a source of authority on specific topics. Here is how to build one.
Jun 20, 2026 -
The case for running your business on fewer, better AI systems
The solopreneurs with the best AI results are not running the most tools. They are running the fewest. Depth beats breadth in AI operations. Here is why, and how to get there.
Jun 19, 2026 -
The solopreneur's guide to AI-powered project management
Project management for a solopreneur is mostly overhead: status updates, task tracking, deadline monitoring, and client communication. AI handles all four. Here is the setup that returns five hours per week.
Jun 19, 2026 -
What Perplexity actually looks for when it cites a source
Perplexity cites sources differently from ChatGPT and Google AI Mode. It re-crawls frequently, weights freshness heavily, and prefers pages with extractable, standalone answers. Here is what that means for your content strategy.
Jun 18, 2026 -
How to use AI to onboard clients faster and retain them longer
Client onboarding is the function that sets the tone for the entire engagement. A slow, manual onboarding signals disorganization. An AI-assisted onboarding that delivers a personalized brief within 24 hours signals competence. Here is the setup.
Jun 18, 2026 -
The AI tools worth paying for in 2026 (and the ones that are not)
Most solopreneurs pay for more AI tools than they use. This is an honest assessment of which categories are worth the subscription and which are not — based on 18 months of running a one-person agentic operation.
Jun 17, 2026 -
How to build an AI-powered newsletter as a solopreneur
A newsletter is the highest-ROI content channel for a solopreneur. It builds a direct audience, drives discovery calls, and compounds over time. AI makes it possible to publish weekly without it consuming your week.
Jun 17, 2026 -
How to get your website cited in Google AI Mode
Google AI Mode synthesizes one answer per query and cites two to four sources. Getting into that citation set requires different signals than traditional Google ranking. Here is what those signals are and how to implement them this week.
Jun 16, 2026 -
The AI solopreneur's guide to pricing and positioning
AI lowers your cost of production. That does not mean you should lower your prices. It means you should raise them — because you are delivering more, faster, with higher consistency. Here is how to position and price an AI-augmented solopreneur practice.
Jun 16, 2026 -
How agentic AI is changing what it means to be a founder
The founder role is changing. Not because AI is replacing founders — but because the work of founding is shifting from building and doing to designing and governing. The founders who adapt to that shift will operate at a scale that was previously impossible alone.
Jun 15, 2026 -
Why AI-generated content fails (and the three fixes that work)
Most AI content fails for three reasons: a vague brief, no brand voice constraint, and no evaluation step. None of these are model problems. They are system problems. Here is how to fix each one.
Jun 15, 2026 -
The AI content calendar: how to plan a month of posts in two hours
A content calendar is not a publishing schedule. It is a brief queue. AI fills the queue in two hours. You spend the rest of the month editing, not planning.
Jun 14, 2026 -
How to use AI to run your finances as a solopreneur
Bookkeeping, invoicing, cash flow forecasting, and tax prep summaries — four financial functions that consume hours of a solopreneur's week. AI handles the reading and the summarizing. You handle the decisions.
Jun 14, 2026 -
What the best AI solopreneurs have in common
After 18 months of running an agentic one-person operation and working with solopreneurs who are doing the same, a pattern is clear. The ones getting the best results share five traits. None of them are technical.
Jun 13, 2026 -
The solopreneur's guide to AI-powered customer support
Customer support is the function that scales worst for solopreneurs. Every new client adds support volume. AI triage handles the first 80% of that volume without you. Here is the setup.
Jun 13, 2026 -
How to measure whether your AI is actually working
Most solopreneurs cannot answer the question: is my AI actually working? They know they are using it. They do not know if it is producing better outcomes. Here is the measurement framework.
Jun 12, 2026 -
AI and the solopreneur's biggest fear: losing the human touch
The fear that AI will make your work feel impersonal is legitimate. It is also solvable. The solopreneurs who use AI most effectively are the ones whose work feels more personal, not less — because AI handles the volume and they spend more time on the relationship.
Jun 12, 2026 -
The compounding advantage: why starting your AI stack now matters
An AI stack that has been running for 12 months is not 12 months better than one that just started. It is exponentially better. The prompt library, the evaluation data, the refined workflows — these compound. Starting now is not optional.
Jun 11, 2026 -
How to write a brand brief that makes AI sound like you
A brand brief is the document that tells AI who you are, how you write, and what you never say. Without it, every AI output sounds like the internet average. With it, AI sounds like you. Here is how to write one in 45 minutes.
Jun 11, 2026 -
AI for solopreneurs: the honest guide to what works and what does not
Most AI content for solopreneurs is written by people who have not run a one-person business. This is written by someone who has. Here is what actually works, what does not, and what the hype gets wrong.
Jun 10, 2026 -
How to use AI to deliver better client work in less time
The solopreneurs using AI most effectively are not using it to do the same work faster. They are using it to do better work — more research, more drafts, more iterations — in the same time. Here is the pattern.
Jun 10, 2026 -
The future of work is not AI replacing you. It is AI reporting to you
The replacement narrative is wrong. AI does not replace the solopreneur — it reports to them. The shift is from doing the work to governing the system that does the work. That is a management skill, not a technical one.
Jun 9, 2026 -
From overwhelmed to operational: a solopreneur's 90-day AI reset
If you have been trying to get AI working for six months and still feel behind, the problem is not the tools. It is the approach. Here is a 90-day reset that moves you from overwhelmed to operational — one function at a time.
Jun 9, 2026 -
The GEO audit: how to check if AI engines can find your content
A GEO audit takes 90 minutes and tells you exactly which of your pages are being cited by AI engines, which are invisible, and what to fix first. Here is the step-by-step process.
Jun 8, 2026 -
How to run a one-person sales operation with AI
A solo sales operation used to mean doing everything manually: research, outreach, follow-up, CRM updates, pipeline reviews. AI handles four of those five. Here is the setup.
Jun 8, 2026 -
Why the next wave of AI will make solopreneurs more competitive, not less
The conventional fear is that AI will favor large companies with large budgets. The data points the other way. The next wave of AI — agentic, autonomous, and cheap — disproportionately benefits the operator who can move fast and govern a system well.
Jun 7, 2026 -
The AI solopreneur's guide to client acquisition
Client acquisition is the function solopreneurs most often leave manual. It is also the function with the clearest AI leverage: lead research, outreach personalization, follow-up sequencing, and proposal drafting are all automatable today.
Jun 7, 2026 -
What solopreneurs get wrong about AI automation
The most common AI automation mistake is automating the wrong thing first. Solopreneurs automate what is interesting, not what is high-volume. The result is a sophisticated system that saves 20 minutes per week.
Jun 6, 2026 -
How to use AI to write content that gets cited, not just ranked
AI-generated content that is not structured for citation is invisible to AI search engines. The same AI that writes your content can also structure it for maximum citation rate — if you give it the right prompt.
Jun 5, 2026 -
The AI skills that will matter in five years (and the ones that will not)
Prompt engineering is already being automated. Model knowledge is a commodity. The AI skills with a five-year shelf life are the ones that sit above the model layer: evaluation, system design, and the judgment to know when AI is wrong.
Jun 5, 2026 -
Stop using AI as a search engine. Start using it as a thinking partner
Most people use AI to look things up. That is the lowest-value use of the technology. The highest-value use is as a thinking partner: a system that challenges your assumptions, stress-tests your plans, and surfaces the thing you have not considered.
Jun 4, 2026 -
How AI changes the economics of a one-person consulting business
A solo consultant's revenue has always been capped by hours. AI breaks that cap — not by making you work faster, but by letting you deliver more without working more. Here is what the new economics look like.
Jun 4, 2026 -
Decision fatigue is the real AI problem. Here is the cure
The AI industry produces more decisions per week than any solopreneur can make well. Which model, which tool, which workflow, which agent. Decision fatigue is not a personal failing — it is a designed-in feature of a market that profits from your attention. Here is how to opt out.
Jun 3, 2026 -
The AI solopreneur stack for 2026: what we actually run
This is the actual stack Avakata runs as a one-person operation: the tools, the agents, the integrations, and the weekly time investment. No affiliate links, no sponsored mentions. Just what works.
Jun 3, 2026 -
Why your AI outputs sound generic (and how to fix it in one prompt)
Generic AI output is not a model problem. It is a prompt problem. The model produces the average of its training data unless you give it a specific voice, a specific audience, and a specific constraint. Here is the one prompt addition that fixes it.
Jun 2, 2026 -
Running a one-person marketing department with AI
A full marketing department — content, SEO, paid media, email, social, analytics — used to require six people. In 2026, a solopreneur with the right AI stack runs all six functions. Here is the exact setup.
Jun 2, 2026 -
How to pick your first AI agent without wasting a month
Most people spend four weeks researching AI agents and deploy nothing. The decision framework is simpler than the research suggests: pick the task with the highest volume, the clearest output standard, and the lowest risk if it goes wrong.
Jun 1, 2026 -
AI search is replacing Google for 40% of queries. Here is what to do
Gartner projects a 25% drop in traditional search volume by 2026. ChatGPT, Perplexity, and Google AI Mode are handling queries that used to go to the blue links. If your content is not optimized for AI citation, you are invisible to a growing share of your audience.
Jun 1, 2026 -
The solopreneur content machine: publish daily without burning out
Publishing daily as a solopreneur is not a willpower problem. It is a system problem. The solopreneurs who publish consistently are not working harder — they have a content machine that runs on a brief, not a blank page.
May 31, 2026 -
Your AI stack is only as good as your evaluation layer
Most AI failures in production are not generation failures. They are evaluation failures. The model produced something. Nobody checked if it was good. Here is how to build the evaluation layer that makes your AI stack trustworthy.
May 31, 2026 -
Agentic AI in 2026: what it means for your business right now
Agentic AI is not a feature upgrade. It is a shift in who initiates work. When AI moves from responding to your prompts to acting on its own triggers, the operating model of a business changes. Here is what that shift means in practice.
May 30, 2026 -
How to build a prompt library that actually saves you time
A prompt library is not a collection of clever prompts. It is a system of reusable inputs that produce consistent, on-brand outputs without you rewriting the brief every time. Here is how to build one that compounds.
May 30, 2026 -
The shiny object trap: why most solopreneurs never get AI working
The solopreneurs who get the least from AI are the ones who follow it most closely. Constant tool-switching, demo-watching, and newsletter-reading is a full-time job that produces zero output. Here is the pattern and how to break it.
May 29, 2026 -
GEO for solopreneurs: how to get cited by ChatGPT and Perplexity
ChatGPT, Perplexity, and Google AI Mode do not return a list of links. They return one synthesized answer and choose whose words to use. Getting cited in that answer is the new SEO — and solopreneurs have a structural advantage if they know how to use it.
May 29, 2026 -
Five business functions a solopreneur can hand to AI today
Content, support, lead qualification, scheduling, and bookkeeping summaries. These five functions consumed the majority of a solopreneur's week in 2023. In 2026, each one has a working AI solution you can deploy this week.
May 28, 2026 -
The AI future of business is already here. Most companies are not in it
The companies using AI as a competitive weapon are not the ones with the biggest budgets. They are the ones who stopped treating AI as a tool and started treating it as infrastructure. The gap between them and everyone else is widening every quarter.
May 28, 2026 -
How to make AI work for you in the next 30 days
Most people spend 30 days evaluating AI. The ones who get results spend 30 days shipping with it. Here is a day-by-day framework for moving from curious to operational in one month.
May 27, 2026 -
What AI agents actually do all day
Most people picture AI agents as chatbots with extra steps. The reality is narrower and more useful: a well-designed agent watches one signal, makes one decision, and executes one action — on a loop, without you.
May 27, 2026 -
AI overload is real. Here is how to stop drowning in it
There are now more AI tools than any person can evaluate. The answer is not to evaluate more of them. It is to stop evaluating and start shipping with the three you already have.
May 26, 2026 -
The one-person business is no longer a compromise
Running a business alone used to mean doing less. In 2026, a solopreneur with the right AI stack operates at the output of a 10-person team — without the payroll, the meetings, or the management overhead.
May 26, 2026 -
Why citation is the new ranking
For thirty years SEO optimized for a position on a ranked list. ChatGPT, Perplexity, and Google AI Overviews do not return lists — they return one synthesized answer and silently choose whose words to cite. Here is how generative engine optimization (GEO) actually works, and what six months of client audit data shows moves the needle.
May 25, 2026 -
The case for boring agents
The AI agents that ship the most value in production are also the most boring — predictable, narrow, observable, and easy to roll back. Here is why we build for boredom on purpose, and how to design agents that are dependable instead of impressive.
May 24, 2026 -
14 changes shipped this week. Here is the diff
An auto-published engine log: 14 changes shipped to this site in seven days — twelve copy rewrites, one new FAQ, one image swap — for a net +1.4 point conversion lift. The full diff, why each change was made, and what the critic rejected.
May 23, 2026 -
Freshness stamps are a ranking signal for LLMs
Pages with a visible "last updated" date get cited roughly 2.3x more often by Perplexity than pages without one — for the same underlying content. Here is why generative engines treat visible freshness as a trust signal, and how to use it without faking it.
May 19, 2026 -
Five-person agency, one engine
A case note: Aurora Home Goods stopped writing product descriptions by hand in March. The engine wrote 1,412 of them, lifted PDP conversion, and gave a five-person team the output of a department. Here is exactly what happened, with the numbers.
May 14, 2026 -
Why the orchestration graph stays a black box
We are radically transparent about what the engine changes and why — and deliberately opaque about how the 160+ agents are wired together. Here is the honest commercial reason and the honest strategic one, and why opacity about the graph is compatible with transparency about outcomes.
May 9, 2026 -
Three rollbacks. None of them ours
An engine log on reversibility: three changes were rolled back this period. Two were client-initiated — both re-shipped a week later once the data came in — and one was a critic-gate rollback. Zero were forced by a change that broke something. Here is why that distinction matters.
May 2, 2026 -
The critic gate matters more than the writer
In an agentic content system, the evaluation layer — the critic — determines quality far more than the generation layer. It is not the model that writes that makes the system good; it is the model that decides what is allowed to ship. Here is how to build a critic gate that earns trust.
Apr 28, 2026 -
Hiring an in-house AI lead. What to insist on
A short, opinionated checklist of what an in-house AI lead role actually does — and what most job descriptions get wrong. If you are hiring one, here is what to insist on, and the questions that separate operators from prompt hobbyists.
Apr 20, 2026