UnKAnsciousUnKAnscious · Services
Selected engagements where our capability changes the outcome.
UnKAnscious builds software — GoalRunner is a real product. We also undertake selected engagements where the problem is worth senior attention that stays on it.
Discuss an engagementSenior judgment, close to the work.
What is brought to bear
Capability, applied narrowly.
Research, engineering, security, evaluation, automation and model intelligence run against one problem at once. Each one is evidenced below.
The volume
Most of what it produces you never see.
Breadth is only useful if it is filtered. The work that reaches you is the part that survived being examined.
The gate
And almost none of it survives judgment.
Every channel runs into one gate. What continues past it continues because a senior person will stand behind it, and stays accountable for it.
The outcome
One outcome. One accountable name.
Where an engagement needs expertise beyond the senior person leading it, a named specialist is brought in and approved by you before anyone starts.
AI recommends. Humans decide.
- What it readsGmail is the only evidence source GoalRunner reads today.
- What it sendsNothing is sent without you.
- Where it isEarly access · $39/month
Two different things, kept apart on purpose
GoalRunner is a product. You connect your mail, it tells you what you owe and what you are owed, and you pay for software. It is early-access software at $39 a month. If that is what you came for, the GoalRunner pages are the right place and this one is not.
Services is a practice. It is work performed for you — scoped, priced and staffed per engagement, with the people named before it starts. Buying one has never been a condition of the other, and it never will be.
Six capabilities, each one evidenced.
Each of these is evidenced in the work, and the evidence is named. A capability we cannot cite is not published.
AI orchestration and model routing
Choosing and governing which model does what, under an explicit authority and cost policy.
model_router — router, authority, policy and economics, with a machine-readable capability registry and maturity rungs.
Agentic engineering
Building systems that reason over real evidence and act only where a person permits.
GoalRunner itself — the engine pipeline, the shipped client, and 798 test files.
Security architecture and governance
Designing the authority boundary, and the reporting that survives being questioned.
An incident response plan, five remediation records, a tabletop, and a CASA AL1 assessment in progress — no letter of validation has issued; and the published record on Bron Torres.
Evaluation and measurement
Deciding whether a change to judgment behaviour is safe to ship, on evidence rather than impression.
A pass/fail release gate with no override path, a golden dataset specification, and precision and recall instruments.
Automation and operations
Running the work on a schedule, with a durable record of what happened and what it cost.
Workers and schedulers, a durable program-state ledger, and invoice-grade billing export.
Research and analysis
Establishing what is actually true about a business before recommending anything.
Business-operations triage, finding and summary, and the written audit record.
Four kinds of engagement
Each one is a problem we have actually worked on, described honestly enough that you can tell whether it is yours.
AI Strategy & Implementation
Deciding what to build with AI, what to buy, and what to refuse — then getting the first real thing into production instead of into a deck.
Typically engaged when
- The organization has agreed it should "do something with AI" and nobody can name the first system.
- A pilot works and nobody can say what would make it safe to run unattended.
- There is a plan to give a model authority over a real workflow — sending, filing, approving, paying — and no one has drawn the authority boundary yet.
What you get
- A written position on what to build, what to buy, and what to leave alone — with the reasoning, not just the recommendation.
- An authority model for any agentic system in scope: what it may understand, what it may do, and what a human must press.
- The evidence design — what the system must be able to show for every conclusion it reaches.
Cybersecurity & Risk Advisory
Security work at the level where it meets the board, the auditor, the regulator and the acquirer — architecture, governance, compliance programs, and the reporting that has to survive being questioned.
Typically engaged when
- A regulated or government environment has to be built, approved, and then actually operated.
- A compliance program exists on paper and does not hold up when someone tests it.
- A transaction requires security due diligence, or an acquired company has to be integrated without inheriting its problems.
What you get
- An assessment that names the real gaps in priority order, with what it costs not to fix each one.
- Security architecture and governance design for the environment in question, including the operating model that keeps it true after the project ends.
- Due-diligence work products for pre-close review and post-close integration.
AI + Security Readiness
The intersection: what happens to your security posture when AI systems are given access to your data and, eventually, permission to act.
Typically engaged when
- Employees are already using AI tools and the organization has no position on what that means for its data.
- A vendor's AI feature is about to be enabled across the company and someone has to say yes or no.
- An agent is going to be given credentials, and the question of what it is permitted to do with them has not been answered.
What you get
- A governance position for AI use that people can actually follow, written in the language of your business rather than a framework's.
- An authority-boundary design for any agent given credentials: least privilege applied to a thing that acts on its own.
- The material you need to answer a customer, an auditor or a board on this subject without improvising.
Executive / Fractional Advisory
A senior technology and security executive available on a standing basis — for organizations that need the judgment without the headcount, and for founders who need someone who has already made the decision in front of them.
Typically engaged when
- You need CISO-level judgment on a recurring basis and cannot justify a full-time hire.
- A founder or an executive team wants a technical advisor who will disagree with them.
- A specific decision — an architecture, a hire, a vendor, a disclosure — needs a second senior opinion before it is made.
What you get
- A standing engagement with defined availability, not an open-ended retainer.
- Direct participation where it is useful: board and customer conversations, architecture review, incident response readiness.
- A clear end. Advisory work that cannot name its own finish line has become a dependency, which is the opposite of the point.
How we work
Selective, and we will say no
We take an engagement when our capability can genuinely change the outcome and the work is worth doing. Selectivity is about fit, impact and trust — not a queue, and we say so early when it is not a fit.
Scoped, not metered
Engagements are scoped to an outcome and priced per engagement. We do not sell hours, and we do not sell a body to sit in your standups.
Written down
Every engagement produces something durable — a position, an architecture, a program someone else can run. If the only artifact is a conversation, you paid for the wrong thing.
Who does the work
Senior leaders stay close to the work. The person who scopes an engagement is the person accountable for what it produces. AI-native execution expands what that leadership can research, build, secure, analyse and deliver — without putting layers between judgment and the outcome.
Engagements are currently led by Bron Torres, co-founder and co-CEO — twenty years of enterprise security leadership, and the architect of GoalRunner's engine, orchestration and governance controls. Where an engagement needs expertise beyond the person leading it, a named specialist is brought in and approved by you before anyone starts.
The full record — roles, dates, programs and speaking — is on the Bron Torres page.
What we will tell you before you ask
- If your procurement process requires a vendor of a particular size or tenure, that is a real constraint and we would rather you know it now than discover it at contracting.
- We hold no security certifications. UnKAnscious is not SOC 2 audited, ISO certified, or FedRAMP authorized as a company. Experience taking other organizations through those regimes is not the same thing as holding them ourselves, and we will not let the two blur.
- We are not a law firm or an audit firm. Nothing in an engagement is legal advice or an audit opinion, and we will tell you when you need one.
- We do not publish client names or results. Not modesty — we do not have permission to, and we would rather have a short page than a fabricated one. Ask about references and we will tell you honestly what we can and cannot offer.
Questions about engagements
What does an engagement cost?
It depends on the scope, and we would rather tell you a real number for your actual problem than publish a range that is wrong for everyone. Scope and price are agreed in writing before any work starts.
Do you have professional liability or cyber insurance?
Ask during scoping and we will tell you exactly what is and is not in place at that moment. We will not publish a coverage claim you would then rely on.
Will you sign our NDA, MSA, and security questionnaire?
Yes to the first two as a matter of course. On security questionnaires: we will answer them accurately, including the answers that are "no" — see the disclosures above.
Can you help us evaluate an AI vendor?
Yes, and it is one of the more useful things an outside party can do. The questions worth asking are rarely about model quality; they are about what the system reaches, what it retains, what it may do without a human, and what it can show you afterwards.
We are looking for the product, not a consultant.
Then you want GoalRunner, and the fastest route is the pricing page. Nothing on this page needs to be part of that.
Discuss an engagement
Tell us the problem in a paragraph. If it is a fit, the next step is a conversation with the person who would lead the work. If it is not, we will say so and, where we can, point you at someone better placed.