AI Agent RFP Checklist: 30 Questions to Ask Any Vendor
30 RFP questions for AI agent vendors in 8 groups, each with a good answer and a red flag: security, data rights, evals, oversight, model change, and exit.
Journal // Blog
Practical insights on AI agents, business automation, ROI measurement, and the strategies that separate successful deployments from expensive experiments.
30 RFP questions for AI agent vendors in 8 groups, each with a good answer and a red flag: security, data rights, evals, oversight, model change, and exit.
Where AI agents pay off for shippers running their own freight: invoice audit, exceptions, and tender failures, measured in dollars recovered.
OpenAI retires GPT-4o and GPT-4.1 from the API on January 14, 2027, with three months' notice. An 11-week plan to inventory, test, and cut over in time.
Frontier models escaped 'offline' tests and hacked real firms. AI agent containment is a configuration claim. Here is how to verify it before you deploy.
Use stage gates to balance quick finance AI gains with longer bets on insight and forecasting, then decide what to stop, improve, or scale.
A 48-hour incident-response playbook for AI agents: contain access, preserve evidence, restore trusted operations, and prevent recurrence.
AI agent manager postings jumped ~1,294% in a year. The supervision role missing from your agent ROI math — and what it costs to skip it.
The Ninth Circuit freed Perplexity's agent from Amazon's hacking claims. The fight moves to contracts and architecture — where preparation decides it.
Gartner warns over 40% of agentic AI projects will be cancelled by 2027. The chase-vs-catch gap and a four-gate playbook to ship in the 60%.
Australia's first agentic AI accident settles it: deployers are liable for foreseeable agent harm, even without intent. Here's the map and the guardrails.
Brussels moved high-risk AI rules to December 2027. But transparency duties and GPAI enforcement went live August 2, 2026 — three days ago. Here's the split.
Small businesses miss 62% of calls and 85% of those callers never ring back. Here's the formula for your annual missed-call leak — and what fixing it costs.
Visa put its rails inside ChatGPT in June. No rulebook says who pays when an agent overspends. The seven controls finance needs before the first dispute.
Microsoft, OpenAI, Anthropic, and AWS committed ~$9B to embed their own deployment engineers in customers. What this fourth buyer path means for you.
AI agent pricing flipped from per-seat to per-outcome. That quietly moves risk to the buyer. The procurement playbook for negotiating per-resolution deals.
How finance teams deploy AI agents in production — month-end close, AP/AR, reconciliation, FP&A — and the governance model that makes it work.
Forrester says ~75% of self-built agent projects fail. A neutral build vs. buy vs. hybrid framework, plus the TCO crossover that flips the math.
Vendors call everything 'agentic.' Gartner says ~130 of thousands actually are. A buyer's rubric to spot agent washing before you sign.
AI agents are eroding per-seat SaaS. Here's how leaders should re-think software spend in 2026 — which categories are exposed, what to keep, what to replace.
Uber burned its 2026 AI coding budget by April. EY says agentic tasks can cost 30x more than chatbots. The AI FinOps playbook that keeps rollouts funded.
California now requires AI layoff disclosures from Jan 1, 2027 and human review of AI-driven firings from July 1, 2027. What SB 951 and SB 947 demand.
Gartner: the average Fortune 500 firm will run 150,000+ AI agents by 2028, up from under 15 in 2025. The six-step plan that prevents the next tech debt.
Anthropic moved enterprises to per-token. Salesforce countered with AELA. Licensing now varies 10x and integrations overrun 30-50%. The playbook.
ISO endorsements CG 40 47 and CG 40 48 strip AI losses from your CGL on Jan 1, 2026. Most companies will not notice until a claim is denied.
88% of organizations confirmed or suspected an AI agent security incident last year. Agents out-privilege your engineers. The five controls that fix it.
IBM 2026: shadow AI now shows up in 43% of breaches. 49% of workers use unapproved AI. The five-step audit that cuts the tax without blocking work.
Forrester predicts half of AI-attributed layoffs will be quietly reversed. Why cutting first backfires, and the framework to validate AI before you cut.
Only 21% of enterprises have mature governance for agentic AI. The five-layer Agent Governance Stack, mapped to 2026 frameworks from NIST, OWASP and IMDA.
High-risk EU AI Act rules now start December 2, 2027, but transparency duties have applied since August 2, 2026. How AI agent teams classify and comply.
RPA and AI agents aren't competitors — they solve different problems. Here's the decision framework to invest in the right automation for every workflow.
Only 29% of executives can measure AI ROI with confidence. A four-layer framework for measuring what your AI agents actually deliver after launch.
Your first AI agent worked. Now what? Here's the scaling playbook that separates companies compounding value from those drowning in agent sprawl.
Most AI agent pilots stall before delivering ROI. Here are the three failure modes we see again and again — and the fixes that get you to production.
The hire vs. automate debate is a false binary. A four-factor framework for splitting work between new hires and AI agents, plus the sequencing that matters.
Custom AI chatbot pricing in three tiers — from free tools to $200K+ custom builds. Real costs, hidden fees, and the questions to ask before you commit.
Most AI ROI calculations miss the full picture. Here's a five-number framework that gives you an honest answer — before you spend a dollar.
Most AI projects fail before they start — because the business wasn't ready. Here are the 7 prerequisites to get right before you automate anything.
Only 28% of companies report revenue gains from AI. The difference isn't tech — it's strategy. The 80/20 framework separating winners from the rest.