Pulse
Notes on running AI in production — what it genuinely speeds up, where it quietly costs you, and how to keep shipping software your own team can still explain.
- 8 min read
EU AI Act, NIST AI RMF, ISO 42001: what actually applies to a small company
Most AI governance writing is aimed at enterprises with a compliance team. Here is the short version for a company without one: four duties already bite, one deadline lands on 2 December 2026, one common habit makes you high-risk, and most of the rest does not apply to you yet.
Read it - 5 min read
Big companies wrote the new security rules. Your company gets the threats, not the budget.
On 28 October a16z published the new enterprise security agenda: faster AI attacks, AI agents on every laptop and a market for AI-native security tools. It was written for large companies, but every threat in it also reaches a 50-person company. Here is what applies to you and what to do in the next 30 days.
Read it - 6 min read
You cannot stop prompt injection. You can limit the damage.
Anyone who writes an email or web page can give orders to an AI agent that reads it. Nobody has a fix. Here is how to limit the damage: three questions, four permission levels and a 30-minute test.
Read it - 2 min read
Five layers that make AI stick: graph, loop, harness, context, prompt
Most AI projects start with prompts and stall after the demo. We work in a different order: map the work, measure it, add guardrails, add your knowledge, then write the instructions. Then the results feed back to the start.
Read it - 5 min read
AI is not digital transformation
Digital transformation moved records into software. AI changes who does the work. Here is how to own it, budget it and measure it.
Read it - 3 min read
What is a Forward Deployed CTO?
A Forward Deployed CTO is a senior technology leader who works inside your team for a set time, makes the technical decisions, and then hands everything over. Here is how it differs from other roles.
Read it - 6 min read
What is a fractional CAIO, and what does it cost?
A fractional Chief AI Officer (CAIO) leads AI in your business for part of the week. Here is what the role covers, how it differs from a CTO, what it costs against a full-time hire, and when to hire full-time.
Read it - 3 min read
A one-page AI policy for your small business
A simple one-page AI policy for small and mid-sized companies: approved tools, data rules, human review, an agent register, spend caps and more. Free to copy.
Read it - 4 min read
Open-source or closed-source AI? How an SME decides
Neither is right for every AI job. Five questions decide it for a small company: data, cost, capability, lock-in and team skills.
Read it - 4 min read
Nobody owns the number: why AI plans fail
AI plans pass through three sets of hands: a strategy deck, a vendor and an integrator. Nobody owns the result. Here is why, and what one embedded owner changes.
Read it - 2 min read
The AI-Native CTO OS: your first 90 days
The CTO OS is a system, not a service list. See what we install in 12 weeks (Embed, Equip, Hand over) and what you own at the end.
Read it - 5 min read
The 5-layer AI stack for a 20–200 person company
An intelligence stack is not one AI tool. It is five layers: knowledge, models, agents, guardrails and people. A company of 20 to 200 people can build all five without a platform team.
Read it - 15 min read
Jev and System One models: fast, typed AI decisions
TypeSafe's Jev does not chat. It answers typed questions about your data, with honest probabilities, in 70–500ms. How it works, where it fails, and four business uses.
Read it - 11 min read
Harness, loop and graph: three layers of an AI agent
Harness, loop and graph are three questions to answer before an AI agent goes live. Learn what each means and how to find the broken layer.
Read it - 4 min read
What a sane agent spend cap looks like
One monthly budget does not protect you. Learn the three caps that do, and how to set them from your own numbers.
Read it - 5 min read
AI-Native SDLC: A Playbook for Small Teams
Ten practices for building software when AI agents write the code, and which ones to start with at pre-MVP, post-MVP and under $1M revenue.
Read it - 3 min read
The bottleneck moved to review
AI made writing code cheap. Reading it is still slow, and few teams plan for that.
Read it
Make your business AI-native, starting this month.
You speak to the senior lead who would do the work. If we are not the right fit, we will say so.