I’m Tyler – CoFounder of Voker, and I’m so bored with being upset via AI hype claims. I guess you’re too.
I studied physics in school, and labored in information science, ML, and analytics till founding Voker. I’m a skeptical particular person via nature (I feel it is the scientist in me) and my intestine response to any era hype is to be cautiously positive till I see issues confirmed out in information.
I felt this fashion about LLMs after they first hit mainstream. I knew that they had actual attainable packages, however used to be additionally nervous in regards to the lofty advertising and marketing buzz they have been getting.
AI as an business has written exams that exact developers are left to money. Promising complete automation, PhD-level intelligence, and best effects. As any person who is skeptical of that narrative, I nonetheless consider brokers can surely ship, however provided that groups are rigorous about measuring efficiency in manufacturing. Each web site or product has Amplitude or PostHog for click on and pageview analytics; an ordinary solution to perceive who is the use of it and the way. Brokers haven’t any an identical, so we constructed Voker.
We’re the Agent Analytics Platform the place you’ll be able to:
– Track your brokers
– Measure their efficiency
– See what customers are asking
– Know for positive brokers are turning in to your customers
– Optimize in keeping with actual information
You put in our SDK, and Voker collects your agent dialog information, mechanically detecting:
– Consumer intents (E book me a resort in Vegas for subsequent Saturday with a poolside view)
– Corrections (No, that room doesn’t have a poolside view!! TRY AGAIN)
– Agent resolutions (Device Consequence: Room Booked… Good fortune!)
Those computerized annotations are the root for construction a holistic view of agent efficiency and consumer conduct in a single analytics platform.
We requested 100+ AI founders, product managers, and agent engineers how they track their brokers in manufacturing and the solution used to be resounding: via combing via person lines (with the occasional evals sprinkled in). All of them reported that they rely on buyer proceedings to inform them when brokers are messing up. We really feel strongly that there’s a 3rd leg of the agent tracking stool lacking – Agent Analytics.
You shouldn’t must look forward to customers to bitch to be informed {that a} contemporary advised exchange is breaking your resort reserving agent, or that the AI finance consultant you constructed is looking the fallacious device to seem up realtime inventory costs.
Seems the antidote to AI hype is unassuming: measure your brokers diligently, then iterate till you get it proper.
Your customers deserve higher AI studies (all of us do)!
Set up the Voker SDK on our loose tier (as much as 2,000 occasions/mo), and get started construction higher brokers nowadays:
https://voker.ai/
