DBIE Bot
Hyperforce Lifecycle Engineering
- DBIE Bot / BaaS (Bot as a service/platform) is a common platform across the teams where it allows easy onboarding of teams for their slack commands and implementing their own functionality.
- It also provides feature and services that teams can leverage and extend upon. This allows teams to focus entirely on their core functionality.
What is it ?
Advantages
-
Eliminate FEDX onboarding, setting up monitoring dashboards, alert systems, and deployments.
-
Simplified Slack integration with pre-configured OAuth scopes.
-
Effortless user onboarding with a one-click Slack workflow.
How easy it is to build a command ?
class Ping(commands.Command):
class Meta:
name = ".ping"
username = "Ping Greeter"
icon_emoji = ":wave:"
description = "A simple healthcheck command"
show_typing = True
def handle(self, message, options):
message.reply(text="Hello 👋, the bot is up and running!")ping.py
hce/dbiebot
pull request
Overview of key features
@command.rbac
class Ping(commands.Command):
class Meta:
name = ".ping"
username = "Ping Greeter"
icon_emoji = ":wave:"
extra_options = {
"auth": {
"teams": [GUS_TEAM_IDS.HCE],
"users": ["kmanchukonda@salesforce.com", "snerati@salesforce.com"],
"message": "Not Authorized to execute this command."
}
}
def handle(self, message, options):
message.reply(text="Hello 👋, the bot is up and running!")RBAC Guardrails
class Ping(commands.Command):
class Meta:
name = ".ping"
def add_arguments(self, parser):
parser.add_argument('--work_id', 'Provide a work item id like W-123456')
def handle(self, message, options):
gus = Connect(
vault.get_team_secret("PMD", "GUS_SERVICE_ACCOUNT_ID"),
vault.get_team_secret("PMD", "GUS_SERVICE_ACCOUNT_PASSWORD"),
)
message.reply(text = gus.work_item(work_id).text)Bringing your own service account / certs
- 🔒 secure multitenancy - path isolation
- seamlessly own and modify your service account secrets yourself
Network Connectivity
aws-esvc1-useast2
Outbound Network Access
Site Bridge
Gus
Gus
Argus
QuantumK
LLM Gateway
Slack
Git soma
UIP
@scheduler.register(interval = 60 * 24) # once a day
def run():
PIESupportMetricCollection().handle()
class PIESupportMetricCollection(commands.Command):
class Meta:
name = ".pie-metrics"
def handle(self, message, options):
# business logic for metric collection
passScheduling Jobs
Mature Slack SDK integration
message.reply(blocks=[
kit.Actions(elements=[
self.approve_button(),
self.reject_button()
]),
self.user_selection(
message.get_user_id()),
self.static_selection()])
Docs
Teams that have currently onboarded and are building their use cases
- PIE APAC - A metric collector for support ticket resolution time on #pie-support.
- PMD - An interface for using the Gus service account to perform operations on work items and change cases.
- HGCO - Utilizing Git Soma APIs to execute pull requests through a Slack interface.
- CCDE - An interface for accessing the CCRDB database in Commerce Cloud.
- AMA - Communicate with the RAG model of db ops GPT and provide a conversational flow for multiple teams knowledge bases.
AMA (Ask me anything)
Leverage Generative AI for DBIE Knowledge bases
Interface
Vector Stores and Refresh
Current state
- fixed knowledge base
-
manually generate embeddings
on local machine - upload them to s3 on a
regular basis
Expected state
- no cap on number of
knowledge bases - auto generate embeddings in
a falcon environment - auto refresh maintaining compliance
LLM Agents
Thank you
DBIE Bot | AMA | LLM Agents
By Kaushik Rishi
