I'm tired of spending so much time sifting through LinkedIn, Zoom, and Apollo to build contact lists.
Let's develop an AI to handle this task at a larger scale!๐ฅ
@AssemblyAI + @TryTaliAI = crazy delicious?
What's better then hitting Assembly's API's for pitch perfect transcriptions? Getting an AI to help you do it in half the time.
Query / debug / problem solve (without having to wait)
@SmithaKolan@mattmakai@jzzaghloul
Github repo: https://t.co/JJlsF1p1us
Weโre working on augmenting Developer support at @trytali and are investigating how to best apply Agents to solve this problem. If youโd like to learn more shoot us a DM.
...
- โBuild itโs houseโ of knowledge and ask more pointed questions to fill in the gaps & develop a robust answer - If it didnโt have an answer, escalate to a human. If it did, answer directly
Learnings:
- Agents are very useful, but not taking our jobs (yet). Weโll be working further to optimize the workflow & tools
- In order to be useful, agents need to aggregate information based on the results of each tool used. If partial information can be learned from...
...
- If it answered, a Zendesk ticket would be setup marking the ticket as answered - If it couldnโt answer, a ticket would be setup assigning it to a human.
For demo purposes, we included:
- A user could email an email address which hit the botโs endpoint
- The bot went through a multi-step process where it iteratively tried tools to get enough information to answer
...
...the foundation of โknowledgeโ about the problem for the bot
- Reason through the query + available info and determine what other data it needed
- Leverage itโs tool set (Searching the web, Discord convoโs or Documentation) to get more information
We leveraged the ReAct Agent framework within @llama_index to create an Orchestrator model that:
- Ingested Documentation & Discord convoโs to embed and do a vector search via @activeloop
- Rerank the most relevant documents via @cohere on the initial query. This set...
Our goal was to mimic how a human would approach the problem:
1. Try to answer based on their own knowledge
2. If they canโt, do a web search to fill in the gaps
3. See if it or a similar question had been answered by the community
4. If they're still stumped, escalate
Autonomous Agents: Are they destined to seize all Support Jobs? ๐ค
This is the question we aimed to answer at the @augment hackathon where we built an Autonomous Support bot leveraging @llama_index.
Demo: https://t.co/k7ZLjuMGrw
Overview / learnings / code below ๐
What if your docs could speak? Tali, our intelligent AI-powered bot, gives them a voice. Instant answers, relevant insights, all from your existing resources. Think of it as a #devrel superhero. Request access to Tali today at https://t.co/rPMivH8WfH #TaliAI
Introducing @TryTaliAI๐. Tali revolutionizes developer support, instantly answering deep technical questions by analyzing your project's docs, code, Discord, YouTube videos & more. Say goodbye to endless waiting for answers!
For projects and companies, Tali serves as the 1st line of defense, managing the influx of developer queries by providing quick, context-aware solutions. It acts as another member of your team, saving time and resources.