Asghar Mirzaie

LangChain’s survey of teams puts the number at 57% for those running AI agents in production over the 2024 to 2026 period. Yet MIT has put forward figures from the same timeframe indicating 95% of generative AI pilots have yet to make a dent on the P&L, and Gartner is forecasting that by 2027 more […]

You will not find a marketing automation project that has failed on account of the software being incapable. More often than not, it is the workflow itself that is at fault, running on top of poor data or an ambiguous process the team never truly put to bed. There is usually a question of ownership, […]

For the small business owner, the issue is seldom an “AI problem”. It is a matter of workflow, and AI is there to address it provided the scope is right. The difference is worth noting: some 80% of AI projects come to nothing, and you will not see a bad model as the cause. More […]

You get the contract over email. Some one will download it, give it a new name and put it in front of legal. After you have waited and followed up, you get back a revision to pass on to finance, and the final version ends up in a shared drive that is not exactly trusted […]

You will find that the majority of teams don’t have an automation problem per se. They have a process issue and their automation is only serving to highlight it. An invoice sits there for a manual touch, a demo request comes in to an inbox, or some Zap goes quiet because a column was renamed […]

You will not find most enterprise workflow automation projects failing in the engine. They fail in the seams: with a legacy system that has the data, an approval sitting in some inbox, or an exception you did not put in your plans. From the outside, adoption appears to be in good health. Some 70 to […]

You can put together the least expensive AI SaaS product in a hurry, and it is the easiest to put out of business: just a chat box of sorts on top of a hosted model. The playbooks from 2024 through 2026 will tell you that an MVP for AI software can be put in place […]

You won’t find the hard part of an ERP AI bot in the model itself. It is in the periphery: the data plumbing, identity and audit concerns, change management, and the thorny issue of who to blame when the bot makes a mistake. The numbers bear this out. Stanford’s 2026 AI Index tells us 88% […]

At your next product meeting you will hear it: “We can put an agent around it, or fine-tune later and let RAG handle the rest.” You will find yourself nodding along with the room, even as you ask yourself if you have just committed to a chatbot, a search engine or some kind of autonomous […]