Your First AI Project Should Cost Under $500: How to Pick and Automate It
A survey of 8,167 business leaders found that 79% expect to spend under $500 on a first AI project, or aren't sure yet. Gartner says only 3% of companies have reached their desired AI cost savings. Here is how to start small and still prove the value.
A new reader survey from The AI Report asked 8,167 business leaders what they expect to spend on their first AI project. The answer was small. Across all five stages of AI maturity, 79% said under $500 or that they aren't sure yet.
That is good news if you run a small or mid-sized business. You don't need a huge budget to begin. But a small budget only helps if the project is picked and measured well.
What the survey actually says
The AI Report grouped respondents by how far along their company is with AI, from just exploring to scaling across the business. Between 31 August and 13 November 2026, the numbers looked like this:
- 79% of all respondents expect to spend under $500 or are not sure yet.
- 84% of companies still exploring are in that group.
- 66% of companies scaling AI across the business are still in that group.
- The share planning to spend more than $5,000 rises at every stage, from 6% of explorers to 18% of those scaling. Even at the top stage, that is fewer than one in five.
One caveat matters. These are expectations for a first project, not total AI spend. The survey also comes from a newsletter's own readers, so treat it as a signal, not a census.
Still, the pattern is clear. Most companies start small, and even the advanced ones rarely start big.
The catch: cheap is not the same as valuable
Gartner's recent research, reported by ITPro, adds a reality check. Only 3% of organisations have achieved their desired cost savings through AI productivity gains. Yet three-quarters of board members still expect AI to deliver value on productivity and costs, and 85% plan to spend more in the coming year.
Gartner's CV Viverito put it bluntly: "organizations will not generate cost savings from AI at this point." The bigger opportunity, he said, is using AI to do things that were not possible before.
There is also a time problem. A separate Gartner study found that only 21% of the time AI frees up goes into more innovative work. Viverito's summary: the struggle is not creating capacity, it is making that capacity usable.
For a business owner, this means the risk is not overspending. The risk is saving ten hours a week and watching those hours disappear into nothing in particular.
How to scope a first project that earns its place
Use four rules.
1. Pick one repetitive job. Choose something that eats hours every week and follows a pattern: sorting inquiries, chasing invoices, summarising meetings, updating a sheet.
2. Use tools you already pay for. Your email, calendar, spreadsheets and chat app already hold the data. Connect them before buying anything new.
3. Write down the baseline first. Note what the task replaces and how many hours it takes now. This is the number that justifies a bigger budget later.
4. Decide where the saved time goes. Gartner's finding says this step is where most companies fail. Name the follow-up work before you automate, such as more sales calls or faster customer replies.
How to automate this with n8n
Here is a concrete example: an AI-assisted inbox triage for incoming customer inquiries. The same pattern works for supplier emails, support requests or booking requests.
- Trigger. Start a new n8n workflow with the Gmail Trigger node, or a Webhook node if inquiries arrive from a website form. A Schedule Trigger also works if you prefer batch processing every hour.
- Clean the input. Use a Set or Edit Fields node to keep only the sender, subject and message body.
- Classify with an LLM. Add an AI Agent or LLM chain node. Give it a short prompt: label the message as sales, support, billing or spam, and extract the key details such as product, urgency and contact info.
- Route. Use a Switch node so each label goes to a different path. Sales goes to your CRM or a Google Sheet, support goes to a team chat channel, billing goes to the accounts person.
- Draft a reply, don't send it. Have the AI write a suggested response and save it as a Gmail draft. A human reviews and sends it. This mirrors something Fortis CEO Greg Cohen said on The AI Report podcast: nothing goes into production without review.
- Log the savings. Add a Google Sheets node that records the date, category and a fixed minutes-saved estimate per message. After a month you will have a real number.
- Report weekly. A second workflow on a Schedule Trigger sums the log and sends you a short summary.
Step 6 is the one most teams skip. It is also the one that turns a small experiment into a budget request with evidence behind it.
Note that an LLM provider may charge per use, so check current pricing for whichever model you choose and test with a small volume first.
What it means for businesses in Nepal
For Nepali SMEs, the survey's message is encouraging: a first AI project does not need a large investment, and most leaders worldwide are starting the same way you can.
The Gartner warning matters just as much. Many Nepali businesses run on WhatsApp, email, spreadsheets and phone calls. That is a good place to start, because the repetitive work is easy to see. But the gain only counts if the saved hours go somewhere useful, like following up with more leads or answering customers faster.
If your business serves clients abroad, this approach also helps you respond across time zones without hiring extra staff. Start with one workflow, measure it for a month, then decide what to build next.
The takeaway
- Most leaders start with a first AI project budget under $500 or are still unsure.
- Cost savings are not automatic: Gartner reports only 3% of companies have reached the savings they wanted.
- Pick one job, use existing tools, measure the baseline, and keep a human in the review loop.
- Track hours saved from day one so the next budget request has evidence.
Sources
- The AI Report, "Most leaders plan to spend under $500 on their first AI project": newsletter.theaireport.ai/p/most-first-ai-projects-start-…
- ITPro, "Just 3% of companies have recorded cost savings with AI" (Gartner research): itpro.com/business/business-strategy/organizations-will-n…