It support · Production

GrabGPT: How Grab built an internal ChatGPT-like tool after a failed support chatbot experiment

The problem

Grab's ML Platform team support channels were overwhelmed with repetitive user inquiries, with on-call engineers spending more time answering questions than building innovative solutions.

First attempt

An initial documentation-fed chatbot built on chatbot-ui and GPT-3.5-turbo failed because the model's token context window could not accommodate the full platform documentation, requiring heavy summarization that only covered a handful of FAQs; embedding search also proved ineffective.

Workflow diagram · grounded in source
1
Support channels overwhelmed
Trigger
Slack channels were flooded with questions and on-call engineers were spending more time answering repetitive queries than building innovative solutions.
▾ source quote
“Slack channels were flooded with questions and our on-call engineers were spending more time answering repetitive queries than building innovative solutions”
2
Documentation chatbot attempt
Ai action
A chatbot was built using chatbot-ui wired to an LLM, fed with over 20,000 words of platform Q&A documentation to handle user queries.
▾ source quote
“I stumbled upon chatbot-ui, a simple yet powerful tool that could be wired up with LLMs. My idea was to feed the chatbot our platform's Q&A documentation (over 20,000 words) and let it handle user queries”
3
Token-limit and embedding failure
Validation
GPT-3.5-turbo's token limit made the approach unscalable and embedding search also did not work well, leading to the idea being abandoned.
▾ source quote
“it was clear that this approach wasn't scalable. I tried with embedding search and it didn't work that well too, so I decided to give up on this idea”
4
GrabGPT built and deployed
Output
GrabGPT was built over a weekend by extending existing frameworks, adding Google login for authentication, and deploying the tool internally.
▾ source quote
“Over a weekend, I extended the existing frameworks, added Google login for authentication, and deployed the tool internally”
5
Rapid company-wide adoption
Feedback loop
GrabGPT reached over 3,000 users with 600 daily active users by month three and eventually spread to almost all Grabbers.
▾ source quote
“Month 3: Over 3000 users, with 600 daily active users”
Reported outcome

GrabGPT became one of the most widely used internal tools at Grab, with 300 users registering on day one, over 3,000 users and 600 daily active users by month three, and eventually almost all Grabbers using it.

Reported metrics
Users registered day 1300
New users day 2600
New users week 1900
Total users month 3over 3000
Show all 5 reported metrics
users registered day 1300
new users day 2600
new users week 1900
total users month 3over 3000
daily active users month 3600
Reported stack
chatbot-uiGPT-3.5-turbocatwalkGrabGPTOpenAIClaudeGemini
◆ Does this fit your context?

Compare to your context

Tell us your scale, team, and constraints. We'll show what changes at your size, what fails at your scale, and whether this case is a fit, needs adaptation, or won't scale to you. Free demo, no signup.

Compare to your context →
~30 seconds · free
Source
https://engineering.grab.com/the-birth-of-grab-gpt
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

GrabGPT became one of the most widely used internal tools at Grab, with 300 users registering on day one, over 3,000 users and 600 daily active users by month three, and eventually almost all Grabbers using it.

What tools did this team use?

chatbot-ui, GPT-3.5-turbo, catwalk, GrabGPT, OpenAI, Claude, Gemini.

What results were reported?

Users registered day 1: 300; New users day 2: 600; New users week 1: 900; Total users month 3: over 3000 (source-reported, not independently verified).

What failed first in this deployment?

An initial documentation-fed chatbot built on chatbot-ui and GPT-3.5-turbo failed because the model's token context window could not accommodate the full platform documentation, requiring heavy summarization that only…

How is this it support AI workflow structured?

Support channels overwhelmed → Documentation chatbot attempt → Token-limit and embedding failure → GrabGPT built and deployed → Rapid company-wide adoption.

Related it support cases

WHAT TO DO WITH THIS

Now compare it to your context

This case is one data point. Whether its pattern fits you depends on your volumes, your stack, and your exception load — that comparison is the step no case study can do for you.

Browse the It support cases →