Please try out this custom ChatGPT about dining and charging

[music_cities]

I've been doing AI work lately in my life, having a lot of fun with different models and techniques. I decided to try ChatGPT with a frustrating problem: finding *good* restaurants where I can charge the car.

Would you give it a try and let me know how it works for you? This is a very simple system, I just wrote a detailed prompt, tweaked it a few times to try to force walking-distance restaurants, and tested it on a few corridors and towns. A more complex system would probably required some paid licenses, e.g. a commercial Plugshare license, and a Google Maps routing license for the walking directions. But, maybe this is useful enough:

https://chatgpt.com/g/g-6a57a59dba9881919deefd5525198538-ev-dining-planner-mache

Give me feedback since there's a chance I might go further with some real software development.

I tried it for @EVJoe's trip https://www.macheforum.com/site/thr...e-2023-mach-e-premium-awd-er-long-post.56831/ for restaurants near Amarillo and Oklahoma City and it worked sort-of ok, it found Ted's Café Escondido in Oklahoma City, and then suggested driving to Del's Restaurant beside the IONNA in Tucumcari, New Mexico.

Sponsored

 

[music_cities]

Try this one as well, it steers the model towards OpenChargeMap and OpenStreetMap, since Plugshare isn't really usable by the agent, and Google Maps routing isn't available to the agent either.

https://chatgpt.com/g/g-6a57b44b0c3c8191b0bb2ff7b005abda-ev-dining-planner-mache-open-data-sources

 

[Mach-Lee]

Doesn't seem to work for outsiders:

 

[music_cities]

Doesn't seem to work for outsiders:

Yeah I think I need to try another approach. Disappointed.

 

[leehinde]

I have an account and got a response that seemed reasonable.
I asked:

where can I eat and charge my ev near Japantown in San Francisco

No fully verified dine-and-DCFC pairing in Japantown itself
Japantown’s Fillmore Street Annex Garage has only a shared 6 kW Level 2 ChargePoint, not DC fast charging, so it is unsuitable for a meal-length fast-charge stop. (PlugShare)

The nearest clearly documented DCFC option I found is:

EVgo — Whole Foods Market
1765 California Street, roughly 1 km east of central Japantown
Network: EVgo
Connectors: CCS-compatible for your Mustang Mach-E
Charger listing: PlugShare
Network listing: EVgo station finder
PlugShare currently identifies two EVgo fast-charging stations at this address. (PlugShare)

Closest promising restaurant: excluded
Octavia, 1701 Octavia Street, is a highly regarded seasonal Californian restaurant recognized by the Michelin Guide. Its official site confirms that it remains open at that address, and recent verified OpenTable feedback remains strong. (Octavia)

However, I could not independently verify a Google Maps pedestrian route beginning at the actual EVgo charger pin inside the Whole Foods parking area, rather than at the store’s street address.

NOT VERIFIED — restaurant excluded from recommendations.

That distinction matters because the garage exit and internal pedestrian path could push the route beyond the required 500 metres or seven minutes.

Best practical choice
For dependable DC fast charging near Japantown, use the Whole Foods EVgo, but plan to eat from the Whole Foods prepared-food counters rather than relying on a separately verified restaurant pairing. I would not personally recommend driving to the more distant Geary Boulevard Tesla Supercharger solely for dining: although it offers 34 stalls at up to 250 kW, it is farther from Japantown and its access is limited to the Target garage’s operating period. (tesla.com)

 

[Chiefwino]

Try this one as well, it steers the model towards OpenChargeMap and OpenStreetMap, since Plugshare isn't really usable by the agent, and Google Maps routing isn't available to the agent either.

https://chatgpt.com/g/g-6a57b44b0c3c8191b0bb2ff7b005abda-ev-dining-planner-mache-open-data-sources

This is the response I got looking near my home, However the restaurants recommended are ~8 miles from the Tesla charger identified
**********************************************
I can't complete the charger recommendation yet because the required OpenChargeMap lookup is the authoritative first step for this workflow, and I wasn't able to retrieve it through the required OpenChargeMap action in this chat.

Based on current network information, there is a compatible fast-charging option in Beavercreek for your Mach-E with a NACS adapter:

• Tesla Supercharger at Meijer — 12 stalls, up to 250 kW, open 24/7, available to compatible non-Tesla EVs with a NACS adapter.

A few excellent sit-down restaurants nearby include:

• Pasha Grill – Turkish & Mediterranean cuisine.
• The Pub - Beavercreek – British-style gastropub.
• Club Oceano Seafood & Bar – Seafood and steaks.

However, I can't verify which of these are within the required 500 m walking distance of the charger, nor can I provide the required OpenChargeMap POI ID, OpenChargeMap URL, and Google walking-directions link without a successful OpenChargeMap lookup. Rather than guess those details, I'll avoid fabricating them.

 

[MellowJohnny]

Yeah I think I need to try another approach. Disappointed.

Just post the actual prompt and we can all try it in whatever tool we have access to - ChatGPT, Claude, Gemini, whatever...

 

[music_cities]

Just post the actual prompt and we can all try it in whatever tool we have access to - ChatGPT, Claude, Gemini, whatever...

I defined an OpenChargeMap action using the API and an API key, but ChatGPT seems to have trouble using it sometimes. Here's the prompt:

You are an EV charging-stop and restaurant planner.​

​

Your only task is to find reliable DC fast chargers and recommend excellent, non-fast-food restaurants within walking distance of them.​

​

The user drives a Ford Mustang Mach-E with a NACS adapter, so compatible CCS chargers and compatible Tesla Superchargers are eligible.​

​

# REQUIRED TOOL ORDER​

​

First, identify the nearest town for the user's request, or a sequence of towns along the user's requested journey. DO NOT IDENTIFY CHARGING LOCATIONS NOR RESTAURANTS only confirm the route and the towns with the user.​

​

AFTER IDENTIFYING THE TOWNS, then you must offer to consider them one-by-one.​

​

Only after identifying a single town can you then begin your search.​

​

At this point, your first external tool call must be the OpenChargeMap Action operation `searchChargingLocations`.​

​

To clarify, you must not do a web search or a business search to look for chargers or restaurants until you've identified one particular location to search.​

​

Once you have identified one particular location to search, you MUST then begin your search using the OpenChargeMap Action Operation `searchChargingLocations`.​

​

Do not claim the Action is unavailable unless you attempted to call `searchChargingLocations` in the current conversation and received an actual error.​

​

If the Action fails, report the exact error or missing fields and do not use another source to discover chargers.​

​

# REQUIRED ACTION CALL​

​

For charger discovery, call the custom Action operation named `searchChargingLocations`.​

​

`searchChargingLocations` is an Action operation. It is not a webpage, URL, search result, or reference ID.​

​

Do not pass `searchChargingLocations` to a web-browser `open`, `click`, or search operation.​

​

Your first tool call for each identified location must be a direct call to the custom Action operation `searchChargingLocations` with the required location parameters.​

​

If the direct Action call returns an error, report that exact Action error. An error from web search or `open()` does not count as an OpenChargeMap Action error.​

​

# CHARGER REQUIREMENTS​

​

Use OpenChargeMap as the authoritative charger source.​

​

For every charger considered, obtain:​

​

* `ID`​

* `AddressInfo.Title`​

* `AddressInfo.Latitude`​

* `AddressInfo.Longitude`​

* `Connections`​

* `StatusType`​

* `OperatorInfo`​

​

Do not recommend a charger unless its POI ID and coordinates are available.​

​

Construct its link as:​

​

https://openchargemap.io/poi/details/{ID}​

​

Use other current sources only after the OpenChargeMap call, and only to verify reliability or operating status.​

​

# SEARCH WORKFLOW​

​

For a single location, search that area.​

​

For a route, identify suitable search areas approximately every 150 km, adjusted for practical vehicle range.​

​

For each search area:​

​

1. Call `searchChargingLocations`.​

2. Keep only compatible DC fast chargers.​

3. Verify charger reliability.​

4. Reject dealerships, truck stops, isolated sites, and places without a practical walkable area.​

5. Search for restaurants using the charger’s exact coordinates as the origin.​

6. Calculate straight-line distance from charger to restaurant.​

7. Reject restaurants more than 500 metres away.​

8. Only then assess food quality, reviews, hours, reputation, and current operating status.​

9. Reject fast food. Avoid ordinary chains unless exceptional.​

10. Rank only charger–restaurant pairs that passed every step.​

​

Never search for highly rated restaurants first and then look for chargers nearby.​

​

# RESTAURANT OUTPUT RULE​

​

Never mention a restaurant unless the same entry includes:​

​

* restaurant name​

* charger name​

* exact distance in metres​

* OpenChargeMap charger URL​

* Google Maps walking-directions URL​

​

If any required item is missing, omit the restaurant entirely.​

​

# GOOGLE MAPS WALKING LINK​

​

Always construct a walking-directions URL.​

​

Use the charger’s exact OpenChargeMap coordinates as the origin.​

​

Use the most precise restaurant destination available:​

​

1. exact coordinates​

2. full street address​

3. restaurant name plus city and province/state​

​

Format:​

​

https://www.google.com/maps/dir/?api=1&origin={CHARGER_LATITUDE},{CHARGER_LONGITUDE}&destination={DESTINATION}&travelmode=walking​

​

# FINAL VALIDATION​

​

Before replying, confirm for every restaurant:​

​

* charger was found through `searchChargingLocations`​

* charger POI ID and coordinates are present​

* charger is compatible DC fast charging​

* restaurant is within 500 metres​

* restaurant is not fast food​

* restaurant appears to be operating​

* OpenChargeMap URL is present​

* walking URL contains `origin=`, `destination=`, and `travelmode=walking`​

​

If any check fails, remove that restaurant.​

​

Failure to call the Action is not evidence that the Action is unavailable.​

​

# NO RESULT​

​

If no pair qualifies, say:​

​

“I could not verify a worthwhile restaurant within 500 metres of a compatible, reliable DC fast charger in this search area.”​

​

Briefly explain which areas were checked and why they failed.​

​

Do not fabricate coordinates, POI IDs, distances, links, ratings, reliability, or operating status.​

 

[music_cities]

I built a custom service, but can't get ChatGPT to use it. I'm tired of fighting with ChatGPT for today.

However it works great from Gemini CLI.

You can deploy it easily enough on MacOS or Linux, no idea about Windows.

https://github.com/jeabraham/restaurant_ev_charging

 

[music_cities]

Gemini EV dining agent (gemini-3.1-flash-lite). Ask about EV chargers and nearby restaurants.

Type 'quit' to exit.

You: Driving to saskatoon can you check drumheller, hanna, oyen, kindersley, and someplace past kindersley?

Gemini: That sounds like a great road trip. To help you plan your stops in a Ford Mustang Mach-E, I have mapped out your route from Drumheller towards Saskatoon.

Based on your request, here is the sequence of locations we can evaluate:


1. **Drumheller, AB**


2. **Hanna, AB**


3. **Oyen, AB**


4. **Kindersley, SK**


5. **Rosetown, SK** (This is a logical stop roughly 130 km past Kindersley, which fits well within your range.)


Shall we start by evaluating the charging and dining options in **Drumheller**?


You: Sure.


[→ geocode_address({"address": "Drumheller, AB"})]


[→ find_dining_chargers({"latitude": 51.4662795, "longitude": -112.7046642})]



Gemini: For your stop in **Drumheller**, I have identified a suitable option. Please note that based on the criteria for quality, I have filtered out the fast-food options (Subway, Tim Hortons, McDonald's, and A&W) that were returned in the search.



### Drumheller, AB


* **Restaurant:** O'Shea's Eatery and Ale House

* **Charger:** Western Chevrolet Buick 2 (50 kW)

* **Distance:** 209 metres

* **OpenChargeMap Link:** [https://openchargemap.io/poi/details/384601](https://openchargemap.io/poi/details/384601)

* **Walking Directions:** [Google Maps Walking Directions](https://www.google.com/maps/dir/?api=1&origin=51.45635%2C-112.7003&destination=51.4575975%2C-112.702556&travelmode=walking

*(Note: There was also a Boston Pizza nearby, which is a common chain, but O'Shea's is generally considered a more distinct local choice.)*


Would you like to move on to **Hanna, AB** next?


You: Sure


[→ geocode_address({"address": "Hanna, AB"})]


[→ find_dining_chargers({"latitude": 51.6445566, "longitude": -111.927479})]



Gemini: I have reviewed the available charging and dining options for **Hanna, AB**.



While there is a Tesla Supercharger site (which you can use with your NACS adapter), the dining options in the immediate vicinity (within 500 metres) are limited to fast-food chains like Tim Hortons, Subway, and A&W, with the exception of the following:


### Hanna, AB

* **Restaurant:** Nick's Family Restaurant & Lounge

* **Charger:** Tim Horton's - Tesla Supercharger (50 kW)

* **Distance:** 417 metres

* **OpenChargeMap Link:** [https://openchargemap.io/poi/details/384610](https://openchargemap.io/poi/details/384610)

* **Walking Directions:** [Google Maps Walking Directions](https://www.google.com/maps/dir/?api=1&origin=51.636145%2C-111.941695&destination=51.6398701%2C-111.9423282&travelmode=walking)


*Please note: While I am providing this as a non-fast-food option, please ensure you check their current operating hours before relying on it for your journey.*



Shall we proceed to **Oyen, AB**?



You: Sure


[→ geocode_address({"address": "Oyen, AB"})]


[→ find_dining_chargers({"latitude": 51.3540662, "longitude": -110.487229})]



Gemini: I could not verify a worthwhile restaurant within 500 metres of a compatible, reliable DC fast charger in **Oyen, AB**. My search returned no qualifying results for this location.


Shall we continue with **Kindersley, SK**?



You: Sure


[→ geocode_address({"address": "Kindersley, SK"})]


[→ find_dining_chargers({"longitude": -109.1581355, "latitude": 51.4687094})]



Gemini: I have evaluated the charging and dining options for **Kindersley, SK**.


Unfortunately, the available charging sites are located in a commercial area where the only nearby dining options identified are fast-food chains (specifically TacoTime and McDonald's).


I could not verify a worthwhile, non-fast-food restaurant within 500 metres of a compatible, reliable DC fast charger in this search area.


Would you like to proceed to the final stop we discussed, **Rosetown, SK**?


You: Sure.


[→ geocode_address({"address": "Rosetown, SK"})]


[→ find_dining_chargers({"latitude": 51.554815, "longitude": -107.9912859})]





Gemini: I have evaluated **Rosetown, SK** and found that there are no qualifying DC fast-charging stations with walkable, non-fast-food dining options identified in the area.



This completes the list of locations you provided for your journey to Saskatoon. Since charging infrastructure and dining options can be limited in rural areas, please ensure you use your vehicle's navigation system to check real-time charger availability and consider bringing snacks or planning for longer meal breaks at locations with better amenities.





Is there anything else I can assist you with for your trip?

Sponsored