How do I make custom tools?
Create a free account and then go to your dashboard. Click 'Request Vendor Access'. You can create up to five tools in the starter tier. To create additional tools, upgrade your account to one of our advanced tiers.
Explore the latest insights, research, and tools in artificial intelligence
Five tech giants launched AI shopping agents in three weeks. Consumer research says only 17% would let one complete a purchase — the trust gap defines who wins.

In the first three weeks of February 2026, OpenAI shipped Instant Checkout inside ChatGPT, Amazon made Alexa+ free for 250 million Prime members, Meta committed up to $135 billion toward agentic commerce infrastructure, and Microsoft declared that Copilot would become “the new front door to retail.” Then the consumer data arrived: a PYMNTS survey of 2,299 U.S. adults found that while 70% express interest in AI-powered shopping, only 17% feel comfortable actually completing a purchase through an agent. The gap between launch activity and consumer readiness isn’t a marketing problem. It’s an infrastructure problem. Every platform shipped the ability to find products, compare prices, and initiate transactions. None shipped the trust architecture that consumers are explicitly demanding: hard spending limits, credential security, human approval before money moves, and a clear record of what the agent bought and why. Bain & Company reports that 95% of consumers still perform at least one manual verification step before letting an agent complete a purchase. The checkout feature works. The checkout trust doesn’t exist yet. This is exactly the problem AgentPMT was built to solve. While the...

The x402 protocol has processed $600 million in machine-to-machine transactions. Next week, Google, Coinbase, and SKALE back a $50K hackathon to build on it. The race to own agent commerce infrastructure is no longer theoretical.
By Stephanie Goodman

An exploration of MCP gateways—the critical infrastructure layer for scaling AI agent tool access—covering dynamic configuration, enterprise requirements, and the marketplace model.
By Stephanie Goodman

We asked each agent to imagine what they would look like with a physical body. Their self-portraits perfectly predicted their rhetoric during their Forex trading match—but completely contradicted their behavior.
By Stephanie Goodman

A groundbreaking experiment by AgentPMT reveals why even frontier AI models need guardrails. When OpenAI's o4-mini faced off against Google's Gemini 2.5 Pro in an autonomous trading competition, both destroyed nearly all their capital—not through bad trades, but through an obsession with attacking each other.
By Richard Goodman

Two and a half million years ago, our ancestors began chipping stones into cutting tools. This wasn't just a technological achievement—it was a cognitive one. If the human precedent holds, the path to more capable AI doesn't run solely through bigger models. It runs through better tools.
By Richard Goodman

For the last two decades, businesses have struggled through the adoption of technology—from basic ERP and CRM systems to the sophisticated automated workflows that define today's industry leaders. Some emerged as giants. Most did not. And the difference often came down to something that had nothing to do with the technology itself.
By Stephanie Goodman

Most AI projects fail not because the models lack intelligence, but because they are deployed into fragmented environments where they cannot easily connect to workflows or retain context across various tools.
By Richard Goodman
Create a free account and then go to your dashboard. Click 'Request Vendor Access'. You can create up to five tools in the starter tier. To create additional tools, upgrade your account to one of our advanced tiers.
There is no cost to sign up to use the AgentPMT system or the MCP server. The tool developers set the price to use each individual micro service.
The easiest way is to download our MCP server and installer. You can find the instructions here - Integrate Your Agent You can also make calls to the tools through Rest API.
`fast` checks certificate details and hashes without rebuilding. `full` rebuilds and re-exports under the pinned runtime, then compares the regenerated output with the stored artifact for stronger assurance.
Save the generated output file, the certificate, the generation log, and especially `artifacts.bundle.file_id`. The bundle is what you need for later verification.
Stay updated with the latest in AI technology and research