Simple Prompt Or Agent Workflow? How Not To Overthink Ai

Photobank2/iStock/Getty Images Plus

Follow ZDNET: Add america arsenic a preferred source on Google.

ZDNET's cardinal takeaways

• Gen AI occurrence is astir knowing which attack to use.
• Start pinch nan simplest instrumentality that solves your problem.
• Sometimes you'll request a prompt, and different times an supplier workflow.

"You're astir apt making AI harder than it needs to be." 

This advice from Corey Noles and Grant Harvey's latest section of The Neuron podcast urges greater simplicity successful what has go a analyzable and confusing matter successful caller years -- making generative AI fresh for nan organization. 

Also: 3 smart ways business leaders tin build successful AI strategies - earlier it's excessively late

With 95% of generative AI projects failing, arsenic recovered in a caller MIT study, it's clip to put these initiatives into perspective. 

Many group wrong and extracurricular nan exertion occupation position implementing gen AI arsenic a awesome task that requires rigorous task guidance discipline. Experts moreover mention to that arena pinch ZDNET, telling group not to unreserved into AI, but to deliberation cautiously astir what they are doing first. 

Also: Gartner says adhd AI agents ASAP - aliases else. Oh, and they're besides overhyped

While immoderate AI initiatives, astatine a guidelines level, don't require specified overthinking, others, connected nan different hand, do require much thoughtful planning. So, wherever does a budding AI master tie nan line?

Overall, group "are getting it each wrong" pinch implementing AI, Noles and Harvey asserted. Success is simply a matter of knowing erstwhile nan simplest approaches use -- utilizing a screwdriver alternatively of applying a sledgehammer for a situation that whitethorn require AI. It's besides astir knowing erstwhile much blase approaches are needed to forestall AI projects from going disconnected nan rails.

The prodcasters suggested 4 ways to lick a problem pinch gen AI: 

1. Simply "chatting."
2. Writing a system prompt.
3. Building a project.
4. Deploying an agent. 

Also: Gen AI disillusionment looms, according to Gartner's 2025 Hype Cycle report

"Start pinch nan simplest instrumentality that solves your problem," Noles advised. "Don't build an supplier erstwhile a basal chat will do. We find group get ace excited astir each these awesome tools. But successful practice, you really don't request to jump consecutive to building analyzable agents erstwhile a elemental punctual works." 

But besides "don't walk 2 hours pulling your hairsbreadth crafting nan cleanable punctual and trying to one-shot a analyzable task erstwhile an supplier could grip nan full workflow for you," he added. "There are decidedly times you really do request to trade an epic punctual aliases build an supplier workflow. The instrumentality is knowing erstwhile to usage what."

Also: 5 ways automation tin velocity up your regular workflow - and implementation is easy

Noles and Harvey supply immoderate guidelines to thief professionals determine really overmuch effort an AI-related task needs:

• When to simply chat: "If you want to dive heavy down immoderate rabbit spread astir nan history of bunny rabbits and whether they really eat carrots, you tin perfectly do that successful a normal chat window," said Noles. "You tin do an atrocious batch much location than what astir group would think. There is simply a full batch of worldly you tin do successful a chat model conscionable asking it a mobility without having to do immoderate fancy prompt, without having to get into each of that." 
• When to do a system prompt: A system punctual is champion for a multi-part aliases a formatted output, said Harvey: "A multi-part task would beryllium you request nan AI to spell retired and do aggregate things astatine once: reference aggregate documents, aliases thing much analyzable than a one-off task." Structured prompts are useful for tailored formatting, hyperlinks, and presentation. Structured prompts are besides nan instauration for nan adjacent 2 higher-level categories discussed, projects and agents. 
• When to employment projects: Projects activity champion for repeated-use tasks. "Choose a task erstwhile what you request is simply a predictable result each time," said Noles. This attack useful "where power and transparency matter. The steps are correct there, your instructions are correct there. It tin beryllium updated quickly connected nan fly. If thing isn't working, you tin spell successful and make tweaks really easy without having to log into immoderate different package and propulsion up a node wherever you person to spell successful and make changes." For example, "if you cognize that we alteration really we do reports each 30 days, a task is astir apt correct for you because you tin spell in, propulsion 1 out, driblet different successful -- 20 seconds of your day."
• When to usage agents: Agents are required erstwhile autonomy is portion of nan system. They travel successful to grip complicated, asynchronous tasks that require devices and time. "Use an supplier erstwhile nan task requires autonomy, requires move determination and instrumentality use," said Harvey. "You request nan AI to conscionable spell and fig it retired itself. I can't beryllium babysitting it nan full time; I request it to conscionable do this for me. I request it to beryllium capable to make aggregate decisions connected nan alert connected its own. And I request you to beryllium capable to usage devices for illustration MCP connectors truthful that you tin spell retired and do it."

Gen AI has nan imaginable to beforehand productivity and invention if employed smartly. Understanding nan 4 stages of deployment and nan effort required tin forestall gen AI from becoming a counterforce to nan productivity and invention nan exertion is expected to deliver.