back_office_ops · workflow
Instacart's internal AI assistant Ava (GPT-4) is relied on by 43% of employees to save over an hour per week
Instacart employees across engineering, legal, marketing, and operations were spending significant time on tasks like assimilating unfamiliar codebases, resolving syntax errors, synthesizing meeting notes, and drafting communications, without a secure integrated internal AI tool.
How it works
Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · Employee invokes Ava via Slack
Employees invoke Ava's Slack plugin over 5,000 times per month in threads and group conversations.
Tools used
AvaGPT-4Slack
Outcome
Ava is relied on by 43% of Instacart's company to save over an hour per week, with 60% of engineers generating around 70,000 lines of code monthly using Ava. Daily active users in the Brand Partnerships org surged 50% and daily interactions grew approximately 40% after an internal best-practices roadshow.
Results
Time saved43%
Volumeover an hour a week
Running sinceJuly 2023
Grounding & classification
Source type: technical build writeup
41 fields verified against source quotes.
code generationcontent generationconversational aienterprise searchsummarizationemailknowledge basemeeting recordinghuman review describedmetric backednamed customerproduction runtime claimedtools describedworkflow describedecommerceemployee productivitythroughput increasetime savedtechnical build writeupback office opshr opslegal opsmarketing opsai draft human approvalcase to summary