In the rush to harness artificial intelligence (AI) for business efficiency, some users are pointing fingers at the technology when results fall short.
But what if the real culprit isn’t AI, but the user?
AI integration specialist ET Artificial Intelligence Consultants says that a large proportion of AI implementation failures stem from inadequate task briefing, turning promising tools unproductive.
“AI cannot read the user’s mind, it’s a high-performance team member that thrives on clear direction,” says Keith Levenstein, CEO of ET Artificial Intelligence Consultants.
“Poor management of AI leads to misguided outputs, wasted resources and missed opportunities. It’s important to manage AI like managing any star employee.”
Levenstein says the pitfalls of vague instructions are all too common. A hastily phrased query like ‘Write a marketing plan’ might yield a generic template, ignoring brand specifics or market nuances’. The result? Hours of rework and eroded trust in the technology.
ET Artificial Intelligence Consultants’ latest insights, drawn from its consulting services, underscore that effective AI management begins with precision.
“By treating AI as a collaborative partner, managers can unlock transformative results, boosting productivity,” Levenstein adds.
Properly Briefing AI: The Foundation of Success
“The cornerstone of AI mastery is a robust brief, one that mirrors the structure of a project kickoff meeting. Start with context: What’s the goal? Who’s the audience? What constraints apply? Then layer in specifics: tone, length, format and key elements. End with a call for iteration, inviting refinements,” he notes.
Consider this example, he explains: Instead of asking, ‘Summarise our Q3 sales data’, the briefs should rather be: ‘As our sales director, prepare a 500-word executive summary of Q3 2025 sales data from the attached Excel file. Focus on regional trends in Gauteng and KwaZulu-Natal, highlighting a 15% YoY growth in e-commerce. Use a professional tone suitable for the board meeting on November 20. Include three actionable recommendations, supported by visualisations if possible and cite data sources inline.’
This brief, notes Levenstein, yields a tailored, insightful report, complete with charts and strategies, rather than a bland overview.
ET Artificial Intelligence Consultants recommends testing briefs with a ‘clarity checklist’: Does it answer who, what, when, where, why and how? The payoff?Outputs that align seamlessly with business objectives.
Common Mistakes to Avoid When Briefing AI
Levenstein says even seasoned professionals stumble here. First, ambiguity reigns supreme: Vague prompts like ‘Improve this email’ is open to different interpretations.
The solution is to specify metrics, such as ‘Enhance the subject line open rates by incorporating urgency and personalisation for our millennial audience’.
Second, overloading with jargon confuses AI, which performs best with plain language such as ‘Explain quantum computing simply’ over a technical deep dive brief without context. Third, neglecting iteration: Treating AI as a one-shot wonder ignores its strength in refinement.
Always review and tweak: ‘Revise based on feedback: Add more empathy to my response’. By dodging these traps, managers reclaim control and amplify AI’s potential.
Steering Clear of Hallucinations
AI’s penchant for fabricating plausible-but-false data, known as ‘hallucinations,’ can undermine credibility. A query on market forecasts might spit out invented stats like ‘42% growth in AI adoption by 2026’.
To prevent this, explicitly instruct: ‘Base all figures on verifiable sources from 2024-2025; if data is unavailable, state ‘Data not available’ and suggest alternatives’.
ET Artificial Intelligence Consultants’ protocols include appending ‘Use only empirical evidence’ to prompts, ensuring outputs grounded in reality’.
This not only builds trust but safeguards against regulatory pitfalls in data-sensitive sectors like finance and healthcare.
“Due to the hallucination problems, we specialise in training the AI Agents with personalised information so its answers are only based on information we have supplied,” Levenstein notes.
Demanding Sources for Facts
Transparency is non-negotiable. Always mandate citations: ‘Provide sources for every claim, linking to reputable outlets like Statista or PwC reports’.
This transforms AI from a black box into an accountable ally. For instance, in legal briefings, add: ‘Cite primary sources with URLs or DOIs’.
Setting Rules and Guidelines for Consistent Excellence
Guardrails ensure AI aligns with your ethos. Embed rules upfront: ‘Adhere to our brand voice: Witty, inclusive, and concise, no jargon’. For content creation, specify: ‘Follow AP Style Guide; limit sentences to 25 words’. In customer interactions, rule: ‘Prioritise empathy: Acknowledge concerns before solutions. Examples abound: In HR, ‘Generate interview questions compliant with EEOC guidelines, focusing on skills over demographics’. For marketing, ‘Incorporate SEO best practices: Include keywords naturally, target 1,500 words’.
These directives foster uniformity, reducing compliance risks and enhancing output quality.
Maximising AI for Superior Customer Service
Levenstein says AI shines brightest in customer service, where responsiveness drives loyalty.
ET Artificial Intelligence Consultants’ playbook offers three proven tactics:
- Personalised Query Routing: Brief AI to analyse inquiries: ‘Classify this email by sentiment (positive/neutral/negative) and urgency (low/medium/high), then suggest routing to sales/support/exec. Personalise response starters, eg, Hi [Name], I see you’re excited about our new features, let’s dive in’. This cuts resolution time and provides tailored attention.
- Proactive Issue Resolution: Instruct: ‘Scan chat logs for recurring complaints; generate a FAQ update with empathetic scripts, e.g, We’re sorry for the delay—here’s how we’re fixing it, plus a 10% discount code’.
- Sentiment-Driven Follow-Ups: Prompt: ‘After resolving a ticket, craft a follow-up survey: On a scale of 1-10, how did we do? What can we improve?’If below 7, escalate with a human-touch note’. This loops feedback into continuous improvement, turning one-off interactions into lasting relationships.
“Don’t blame the tool; refine your leadership,” urges Levenstein, who is empowering businesses through bespoke training and integration services.