Why do AI agents in Tea Tree Gully projects pick the wrong tool?
Usually because the tool descriptions do not say clearly what each tool does, so the agent guesses from names. Our live online classes teach learners to write tool descriptions an agent cannot misread. The City of Tea Tree Gully had 101,174 residents in 2021.
Asked how many bus stops serve Tea Tree Gully, our scripted agent answered: none. Not one stop name in South Australia's timetable contains the words "Tea Tree Gully". The agent had searched names because both of its tools were described the same way, as "find stops for a place".
Facts last verified 14 October 2026. Teaching is online; no Tea Tree Gully branch is claimed.
This project needs Python functions, docstrings and an understanding of how agents read tools. These courses build all three; a free trial lesson finds the fit.

Building and directing coding agents, including the tools they call.
See the syllabus
Tool use, function calling and evaluating agent behaviour.
See the syllabus
Making apps with AI help, with clear instructions as a core skill.
See the syllabus
Functions, parameters and docstrings that explain themselves.
See the syllabusBrowse the course atlas for more than one hundred options and use the coding roadmap to check prerequisites.
The four we are known for
These run underneath everything above. Every one is live and online, placed by ability rather than by age, and the first class is free.

Python, web and AI projects where the learner still owns the thinking.
See the syllabus
Automate the work you already do, then let AI carry part of it.
See the syllabus
Train a model, read what it learned, and be able to say why it is wrong.
See the syllabus
Run AI coding agents on real work without losing control of the codebase.
See the syllabusCouncil-wide figures from the 2021 Census for the City of Tea Tree Gully (ABS code LGA47700), set beside South Australia.
| Measure | City of Tea Tree Gully |
|---|---|
| Residents | 101,174 |
| Median age | 41 |
| Families | 29,197 |
| Private dwellings | 41,022 |
| People per household | 2.5 |
| Indicator | Tea Tree Gully | South Australia |
|---|---|---|
| Separate houses | 89.1% (34,400) | 78.0% |
| Workers who took a bus to work | 5.8% (2,944) | 3.1% |
| Workers who drove to work | 65.5% (33,080) | 63.9% |
| Adults with a bachelor degree or higher | 21.0% (17,603) | 22.7% |
Residents aged 5 to 9 numbered 6,030, aged 10 to 14 6,173, and aged 15 to 19 5,846.
2,482 residents were aged 85 or over.
The council area spans about 95 square kilometres in the north-east of Adelaide.
Figures are exactly as the ABS released them; its privacy noise on small cells is one reason we never add them up.
The agent had two Python functions available as tools. stops_named(text) searches stop names. stops_in_council(name) returns stop records whose location falls inside a council boundary. In version one, both docstrings read "Find stops for a place."
| Tool call | Result |
|---|---|
| stops_named("Tea Tree Gully") | 0 stop records |
| stops_named("Tea Tree") | 8 records, all at one interchange |
| stops_in_council("Tea Tree Gully") | 607 stop records |
| Wednesday 21 October 2026 visits at those 607 | 27,084 |
| Tool | Version one | Version two |
|---|---|---|
| stops_named | Find stops for a place. | Search stop NAMES for text. Suburb and council names rarely appear in stop names. |
| stops_in_council | Find stops for a place. | Return every stop record located inside a council boundary. Use for questions about an area. |
With identical descriptions, the agent had nothing to go on but the function names, and "named" sounded like the right match for a place name.
With version two, our scripted agent chose stops_in_council and reported 607 stop records. Nothing else about the agent or the data changed.
An empty result looks like an answer. The agent reported "no stops found" with no error, which is exactly why it is dangerous.
For an agent, a tool's description is the main evidence for when to use it. Write what the tool does, what it does not do, and when to prefer another tool. Test with a question that only one tool can answer correctly, as Tea Tree Gully does here.
Adelaide Metro publishes the timetable openly; working out which stops fall within the council, and all the tallies, was done in our own code. The agent is a small script of ours built to isolate the effect of wording, so no commercial model's output appears here.
A few lines of plain English decide most tool choices.
| Include | Why | In the Tea Tree Gully tools |
|---|---|---|
| What the input means | Agents fill parameters from it | text versus council name |
| What the tool does not do | Prevents near-miss choices | Name search ignores location |
| When to use it | Guides choice between similar tools | Area questions use the boundary tool |
| What an empty result means | Stops false "none found" answers | Zero names is not zero stops |
Label two boxes so clearly that a friend always picks the right one without asking.
Give an AI tool two functions with vague docstrings, ask it a question, then rewrite the docstrings and ask again.
Review the descriptions of tools your agents use at work, especially pairs that sound alike.
Good descriptions help people too. The same docstring that steers an agent tells the next programmer what the function is for.
Council-wide timetable markers for Wednesday 21 October 2026, with weekend days for comparison.
None of the 607 stop records inside the council is a railway station; all visits are by bus, including 243 school bus visits.
66 routes served the council's stop records that Wednesday.
Visits fell to 13,399 on Saturday 24 October and 11,976 on Sunday 25 October.
Council figures
This page covers the whole City of Tea Tree Gully. Suburbs inside it, such as Golden Grove and Modbury, have their own pages, and their figures are not repeated here. Neighbouring councils include Salisbury, Playford, Port Adelaide Enfield, Campbelltown and Adelaide Hills.
The trial lesson matters more than these age bands.
Clear names and instructions in first programs.
Scratch Coding for KidsVibe Coding for KidsFunctions and a realistic idea of AI.
Python and AI for KidsAI Literacy for Kids (Ages 8 to 14)Directing AI tools with precise instructions.
Vibe Coding for TeensAI Tools Course for Teens (13+)Tools, descriptions and agent evaluation.
Generative AI CourseCodex & Claude CodeThe person building the agent. An agent can only be as precise as the descriptions it reads.
Vibe coding often means asking an agent to "add a tool for that". It will, and it may describe the new tool in the vaguest possible words. Later, when two tools sound alike, the agent chooses by guesswork and reports the result with confidence.
Learners who have watched "no stops" become 607 with two rewritten sentences treat tool descriptions as part of the program. That is how they build agents that keep working as more tools are added.
In Tea Tree Gully, the learner who writes clear tool descriptions gets the 607, not the zero. The longer argument is in why teenagers should still learn to code in 2026.
Lessons run live over video, so learners join from home.
Every tool, description and agent answer is reviewed together.
We pause when South Australian schools break and restart with each new term.
One full class without charge, then honest advice.
Grouped by level, or taught one to one.
Two fixed lessons a week, roughly eight a month.
With teachers in India, a 6pm Tea Tree Gully class is 1pm India time in summer and 2pm from April.
Summer and winter time
South Australia's daylight saving runs from the first Sunday in October to the first Sunday in April (IANA time zone database); India has none, so the gap moves between five and four hours.
Three clear prices.
The first lesson is free and ends with a written recommendation.
Group month: around eight lessons with fifteen to twenty learners.
Private month: around eight lessons with one teacher.
Fees are charged in US dollars once the trial has fixed a course and a weekly time. Holiday pauses and missed lessons are covered on the pricing page.
Rated 4.9 across 547 Google reviews. These are real reviews, reproduced as written.
★★★★★
"The one step solution for my son. Modern Age Coders make learning coding so simple that kids love it. The teachers explain complex concepts clearly with practical exercises and interactive content."
Ria Mukherjee
Parent
★★★★★
"Modern Age Coders has been a game-changer for me. I struggled to grasp IT concepts and coding before joining, but their classes transformed everything. I can now confidently write complex programs with ease."
Samriddha Mondal
Student
★★★★★
"One of the most wonderful education centres out there. Education is not limited to school syllabus but focuses on skill development."
Vansh Agarwal
Student
★★★★★
"My child Dhairya is really enjoying the Modern Age Coders classes. This is his first online class and he eagerly looks forward to it. I can already see his improvement, and the teachers are very cooperative."
Sonam Oswal
Parent of Dhairya
★★★★★
"Modern Age Coders have wonderful teachers who teach in a clear, easy and practical way. The teacher boosts students' confidence and inspires them to learn without hesitation."
Sonu Goyal
Parent
★★★★★
"I highly recommend this computer coding class! The teachers are incredibly knowledgeable and passionate about coding."
Ritu Kedia
Parent
A line on the learner's age and hobbies helps us plan. Teens could begin by rewording a pair of tool descriptions, and children by labelling Scratch sprites for a friend.
Questions from across the City of Tea Tree Gully.
The 2021 Census QuickStats count 101,174 residents in the City of Tea Tree Gully local government area, with 41,022 private dwellings and a median age of 41.
A short text, often a function's docstring, telling the agent what a tool does and when to use it. Agents rely on it to choose between tools.
It searched stop names for "Tea Tree Gully" because both tools were described identically. No stop name contains those words, so the search returned zero.
By our boundary check of the Adelaide Metro timetable, 607 stop records lie inside the council, with 27,084 scheduled visits on Wednesday 21 October 2026.
You explain the program you want to an AI assistant, it drafts the code, and you test, question and refine that draft until it works properly.
No. We keep Claude for learners who are 18 or older, and teenagers work with other AI tools.
No. Lessons are live online only, we have no Australian office, and our listed phone number is Indian.
The first lesson is free. Group classes cost USD 100 a month for two live lessons a week, about eight a month, in groups of fifteen to twenty, and one-to-one classes cost USD 150 a month. You agree the course and time before paying.
Fifteen to twenty learners at a similar level and pace, not sorted by age or suburb. One-to-one lessons suit anyone no group time fits.
No. We teach Copilot Studio only in private lessons; groups write agents in Python.
Two suburbs here have their own pages: Golden Grove and Modbury. For the city as a whole, see the Adelaide page and the South Australia guide.