Infrastructure
Monitoring Is a Financial Control
Executives who treat monitoring as an IT purchase are carrying unpriced risk on the balance sheet. Here is how to manage it like every other control.
By Chris Watkins · 13 min read · For CEO, CFO, CTO
Executive Papers
Decisions, argued with evidence.
50 papers for executives on infrastructure, recovery and AI. Each argues one decision, cites its sources (775 in all) and ends with the questions to put to your team.
Disaster Recovery
A Backup Nobody Has Restored Is a Hope
Treat backups as unproven until a scheduled restore test brings a system back intact, and report restore success and restore time to leadership every quarter.
By Chris Watkins · 12 min read · For CEO, CFO
AI
AI Projects Fail on Their Data
Fund discovery, governed data, and reproducible infrastructure before agents, because that is where most stalled enterprise AI projects break.
By Chris Watkins · 12 min read · For CEO, CFO
Start here
For the CFO
For the CTO
Infrastructure
15 papers
Infrastructure
Monitoring Is a Financial Control
Run monitoring like a financial control, with an owner, a target, a defined response, and a monthly review, and watch AI agents as closely as people.
By Chris Watkins · 13 min read · For CEO, CFO, CTO
Infrastructure
If It Lives in One Person's Head, It Is a Risk
Treat key-person dependency in operations as a continuity risk with a price, and retire it with tested runbooks, a second trained person for every critical system, and systems built to be handed back.
By Chris Watkins · 13 min read · For CEO, COO
Infrastructure
Infrastructure as Code Is Business Continuity
Environments that can be rebuilt from a version-controlled repository recover faster, change more safely and pass audits more cheaply than environments rebuilt from memory, so fund infrastructure as code as a continuity control.
By Chris Watkins · 13 min read · For CTO, CFO
Infrastructure
Every Change Should Be Reversible
Require every production change, including configuration and vendor content, to roll out in stages behind health checks that reverse it automatically, so a bad change stays small.
By Chris Watkins · 12 min read · For CTO, Board
Infrastructure
Patch Debt and Shared Passwords Are Balance-Sheet Risk
Treat unpatched known-exploited vulnerabilities, missing MFA, shared or standing credentials, excess privilege, and missing audit trails as measured liabilities with owners and deadlines, because they open the door to the costliest breaches.
By Chris Watkins · 11 min read · For CFO, CEO
Infrastructure
Your Cloud Bill Is an Engineering Problem
Runaway cloud spend comes from architecture and habits, so measure it first, fix it in order of dollar impact, and give every cost a named owner.
By Chris Watkins · 12 min read · For CFO, CTO
Infrastructure
Cyber Insurance Renewal Is a Controls Audit
Run the cyber insurance renewal as a joint CFO and CTO controls audit, because every answer on the application is a warranty that can void cover if the control was not in place on the day of the loss.
By Chris Watkins · 12 min read · For CFO, CTO, CEO
Infrastructure
Kubernetes Is a Commitment, Not a Default
Adopt Kubernetes only when workload count, team size and release frequency justify its operating cost, choose simpler managed runtimes otherwise, and if you adopt it, fund the platform team and the upgrade cycle as a permanent line item.
By Chris Watkins · 11 min read · For CTO, CFO, CEO
Infrastructure
Some Workloads Belong Back On-Premises
Decide where each workload runs with a cost and risk model, placing steady, predictable, data-heavy systems on owned or colocated hardware when the numbers favor it and keeping spiky or fast-changing systems in the cloud, and revisit the placement yearly instead of following an all-cloud or all-on-premises policy.
By Chris Watkins · 11 min read · For CEO, CFO, CTO
Infrastructure
What the Board Should See Every Quarter
Replace ad hoc technology presentations with a fixed quarterly board report of six measures, each with an owner, a target and a trend line, covering restores, exploited vulnerabilities, vendor concentration, change failures, recovery targets and AI systems in production.
By Chris Watkins · 11 min read · For CEO, Board
Infrastructure
Compliance Should Fall Out of Engineering
Keep one control set, check it in code, and collect evidence automatically from the systems that run the business, so that SOC 2, ISO 27001, PCI DSS and customer questionnaires become reports generated from the same data instead of separate manual projects.
By Chris Watkins · 11 min read · For CEO, CFO, CTO
Infrastructure
Business Email Compromise Is a Payments Problem
Treat business email compromise as a payments control owned by finance, with callbacks to numbers already in the vendor master for any change to bank details, independent second approval above a threshold, a hold on first payments to changed accounts, and verification rules that assume voices and video can be faked, backed by email security instead of relying on it.
By Chris Watkins · 12 min read · For CFO, CEO, Controller
Infrastructure
Your VPN Appliance Is the Front Door
Treat internet-facing edge devices as your highest-risk systems by patching them within days, hunting for compromise after every emergency fix, retiring unsupported units, and planning the move from appliance VPNs to identity-based access.
By Chris Watkins · 12 min read · For CEO, CFO, CTO, Board
Infrastructure
The Database Is Your Real Single Point of Failure
Give every business-critical database a named owner and a funded plan covering tested failover, version support dates, capacity and slow-query monitoring, and disciplined schema changes, because the database tier is where outages spread and where data is lost.
By Chris Watkins · 12 min read · For CTO, CEO, CFO, COO
Infrastructure
Identity Is Where Attackers Get In
Treat identity as the primary security control by moving staff and administrators to phishing-resistant MFA, removing standing privilege, inventorying and rotating service accounts, API keys and OAuth tokens, hardening help desk resets, and closing accounts the day people leave, because most breaches now start with stolen or abused credentials.
By Chris Watkins · 12 min read · For CEO, CFO, CISO, Board
Systems Rescue
4 papers
Systems Rescue
Stop the Bleeding Before You Modernize
When production is failing, stabilize and fix root causes in order of risk before funding any rewrite or platform change.
By Chris Watkins · 12 min read · For CEO, CTO
Systems Rescue
Recurring Incidents Are Unpaid Debt
Every incident closed without a root cause returns at a higher cost, so book it as a liability and fund root-cause work as debt repayment.
By Chris Watkins · 12 min read · For CFO, CTO
Systems Rescue
Due Diligence Must Include the Systems
Make technology and cyber due diligence a gating workstream before signing an acquisition, with findings priced into the deal, because the target's systems, debt and breaches become the buyer's liability at closing.
By Chris Watkins · 11 min read · For CEO, CFO, CTO, Board
Systems Rescue
Measure Delivery the Way You Measure Sales
Put five delivery measures in front of leadership every month, per critical service and with agreed targets, and stop judging engineering by headcount, story points or activity.
By Chris Watkins · 10 min read · For CEO, CFO, CTO
Legacy Systems
3 papers
Legacy Systems
Extend Legacy Systems Before You Replace Them
Wrapping, extending and documenting a working legacy system usually beats a rewrite on cost, risk and time, and replacement should happen on a funded retirement plan once it is justified.
By Chris Watkins · 12 min read · For CEO, CFO
Legacy Systems
End-of-Life Software Is an Unpriced Liability
Keep an end-of-support register for every operating system, database, runtime and appliance, and fund retirement or paid extended support before each date, because unsupported software stops receiving security fixes and becomes an audit, insurance and breach liability.
By Chris Watkins · 12 min read · For CFO, CIO, CEO
Legacy Systems
Buy the Commodity, Build the Differentiator
Classify every system as commodity or differentiating, buy and configure the commodity with minimal customization, and spend custom engineering only where it sets you apart.
By Chris Watkins · 11 min read · For CEO, CFO, CTO
Migrations
3 papers
Migrations
A Migration Without a Rollback Path Is a Bet
Every cutover, whether a data center, cloud, database, operating system or framework move, needs a rehearsed rollback, a parallel run and verified data before go-live.
By Chris Watkins · 12 min read · For CTO, COO
Migrations
Your VMware Renewal Is a Migration Decision
Treat the VMware renewal as a platform decision made at least a year before the contract ends, with costed options to renew, renegotiate, move some workloads, or migrate, so you never sign under deadline pressure.
By Chris Watkins · 11 min read · For CEO, CFO, CTO, CIO
Migrations
Every Critical Vendor Needs an Exit Plan
For each critical vendor, keep a documented and tested exit plan that names the data you would take, the alternative you would move to, how long it would take and what it would cost, and review it against early warning signs.
By Chris Watkins · 12 min read · For CEO, CFO, COO, CTO
Disaster Recovery
9 papers
Disaster Recovery
Build High Availability Before Disaster Recovery
Spend first on high availability that absorbs the frequent, local failures, then size disaster recovery to the rare site-wide events that remain.
By Chris Watkins · 12 min read · For CTO, CFO
Disaster Recovery
A Backup Nobody Has Restored Is a Hope
Treat backups as unproven until a scheduled restore test brings a system back intact, and report restore success and restore time to leadership every quarter.
By Chris Watkins · 12 min read · For CEO, CFO
Disaster Recovery
RTO and RPO Are Business Decisions
Have business owners set recovery time and recovery point objectives for each system through an impact analysis, sign them off at executive level, and then engineer and fund to those targets.
By Chris Watkins · 12 min read · For CEO, CFO, COO
Disaster Recovery
Ransomware Readiness Is a Recovery Problem
Whether a company pays a ransom is settled before the attack, by backups the attacker cannot reach, an isolated recovery path, hardened credentials, and a rehearsed clean rebuild.
By Chris Watkins · 11 min read · For CEO, Board
Disaster Recovery
Your Provider's Outage Is Your Outage
Map which cloud, SaaS, network and security vendors your critical services depend on, rank the concentration, and choose a deliberate response for each before their outage becomes yours.
By Chris Watkins · 12 min read · For CEO, Board
Disaster Recovery
Rehearse the Disaster Before It Happens
Tabletop exercises and live failover drills turn a written recovery plan into a capability, so run them on a schedule, track every finding to closure, and report measured recovery times to leadership.
By Chris Watkins · 11 min read · For COO, CTO
Disaster Recovery
The First Hour of a Data Loss Event
Write down now who stops the damage, who freezes writes and backup deletion, who preserves evidence, who owns the restore, who starts the notification clocks, and who chooses the restore point, because the first hour sets the size of the loss.
By Chris Watkins · 13 min read · For CEO, CFO, CTO, COO, General Counsel
Disaster Recovery
Your SaaS Vendor Does Not Back Up Your Data
Back up critical SaaS data, including Microsoft 365, Google Workspace, Salesforce and cloud ERP, to a copy the provider and your own administrators cannot delete, and prove each restore on a schedule.
By Chris Watkins · 12 min read · For CEO, CFO, CTO
Disaster Recovery
Operational Resilience Is Now a Regulation
Map your critical business services, set impact tolerances, and test against severe but plausible scenarios now, because DORA, the UK operational resilience rules, APRA CPS 230 and US interagency guidance require it of financial firms and their contracts carry the same demands to suppliers.
By Chris Watkins · 12 min read · For CEO, CFO, COO, CTO, Board
AI
10 papersAI
AI Projects Fail on Their Data
Fund discovery, governed data, and reproducible infrastructure before agents, because that is where most stalled enterprise AI projects break.
By Chris Watkins · 12 min read · For CEO, CFO
AI
Define Your Data Before You Deploy an Agent
Build a semantic model of ontology, classification, lineage, and access rules for each domain before an agent reads it, because agents without one return confident wrong answers.
By Chris Watkins · 12 min read · For CTO, CDO
AI
AI Agents Need Least Privilege and a Human in the Loop
Treat every AI agent as a privileged insider, giving it its own narrowly scoped identity, logging everything it does, and requiring a named person to approve any destructive or financial action.
By Chris Watkins · 14 min read · For CEO, CTO, Board
AI
Evaluations Are the Test Suite for AI
Treat automated evaluations as the test suite for every AI system, run them on every model, prompt, and data change, block releases that fail them, and keep scoring the system in production.
By Chris Watkins · 13 min read · For CTO
AI
The Real Cost of AI in Production
Inference, embeddings, retrieval and evaluation are a material, variable operating cost, so budget for them explicitly, measure cost per outcome, and choose between cloud APIs and self-hosting on utilization and data constraints.
By Chris Watkins · 12 min read · For CFO
AI
Annual Security Tests Cannot Keep Pace With AI
Replace the once-a-year audit and penetration test with a continuous, standards-based program that scans code and dependencies, fixes known-exploited flaws on a clock, tests exposed systems and AI applications against OWASP, and reports to the board against NIST and ISO.
By Chris Watkins · 12 min read · For CEO, CTO, Board
AI
Prove One AI Use Case in Ninety Days
Answer the board's request for an AI plan by funding one narrow, measurable use case with a baseline, a 90-day deadline and a kill criterion, then scale or stop on the evidence.
By Chris Watkins · 12 min read · For CEO, CFO, Board
AI
MCP Servers Give AI Governed Access
Connect AI assistants and agents to company systems only through a governed integration layer of approved Model Context Protocol servers, each with an identity, scoped permissions, logging and a named owner, instead of letting each team or tool build its own connections.
By Chris Watkins · 12 min read · For CEO, CTO, CISO
AI
Your Employees Already Use AI. Set the Rules.
Publish a one-page AI use policy that names approved enterprise tools, states which classes of data may go into them, makes the employee responsible for checking output, and gives staff a fast way to request new tools, because bans and silence both push use into personal accounts.
By Chris Watkins · 12 min read · For CEO, COO, General Counsel, CISO
AI
Choose Your AI Platform by What You Must Control
Decide between an enterprise AI platform, custom orchestration on model APIs, and privately run open-weight models by naming what the business must control on data, cost at volume, portability, auditability and staff skills, then pick the lightest option that meets those requirements.
By Chris Watkins · 12 min read · For CEO, CTO, CFO, CIO
MSP Partners
1 paper
MSP Partners
Put Senior Engineering Behind Your Brand
MSPs should secure white-label senior engineering capacity for escalations and project work they cannot staff, with written terms for handoff, privileged access and client ownership, instead of declining the work or stretching junior staff.
By Chris Watkins · 12 min read · For CEO, COO, Owner
Discovery
2 papers
Discovery
Map the Systems Before You Fund the Project
Approve no major technology project until a two-to-six-week discovery produces an infrastructure map, a service and dependency map, and a workflow map of what actually runs, and use those maps to set scope, budget and sequence.
By Chris Watkins · 12 min read · For CEO, CFO, CIO, COO
Discovery
Automate the Process You Actually Have
Before funding automation or AI on any workflow, require a measured baseline of how the work is actually done, including exceptions, workarounds and rework, drawn from interviews, observation and system event logs.
By Chris Watkins · 12 min read · For COO, CEO, CFO, CIO
Data
2 papers
Data
Data Classification Is a Breach-Cost Decision
Label every data store by sensitivity, find the shadow copies, and bind retention and access to the label, because breach cost, notification scope and regulatory exposure all scale with data you did not know you held or no longer needed.
By Chris Watkins · 12 min read · For CEO, CFO, General Counsel, CISO
Data
Your Business Needs One Shared Model of Itself
Fund one agreed, business-owned model of the company's core objects, their relationships and the actions allowed on them, mapped to the source systems and used by every report, integration and AI agent, because without it each one invents its own definitions and they disagree.
By Chris Watkins · 12 min read · For CEO, COO, CIO, CDO
Integrations
1 paper
Integrations
Point-to-Point Integrations Are a Hidden Tax
Inventory and price every connection between your systems, then move the critical ones onto a governed integration layer where every feed has an owner, a monitor and a contract.
By Chris Watkins · 12 min read · For CFO, COO, CTO, CIO
No paper fits both filters yet.
From the author
Want the decision made with your numbers?
Chris Watkins wrote these papers from twenty-five years of running production systems. Watkins Labs does the work they describe: stabilization, recovery targets, tested restores, migrations with a way back and AI on a governed foundation.
Ask AI about these papers